diff --git a/crates/polars-core/src/frame/group_by/aggregations/dispatch.rs b/crates/polars-core/src/frame/group_by/aggregations/dispatch.rs
index 6b0258f7ac77..447352a0faaf 100644
--- a/crates/polars-core/src/frame/group_by/aggregations/dispatch.rs
+++ b/crates/polars-core/src/frame/group_by/aggregations/dispatch.rs
@@ -127,6 +127,58 @@ impl Series {
         }
     }
 
+    #[doc(hidden)]
+    pub unsafe fn agg_mean(&self, groups: &GroupsProxy) -> Series {
+        // Prevent a rechunk for every individual group.
+        let s = if groups.len() > 1 {
+            self.rechunk()
+        } else {
+            self.clone()
+        };
+
+        use DataType::*;
+        match s.dtype() {
+            Boolean => s.cast(&Float64).unwrap().agg_mean(groups),
+            Float32 => SeriesWrap(s.f32().unwrap().clone()).agg_mean(groups),
+            Float64 => SeriesWrap(s.f64().unwrap().clone()).agg_mean(groups),
+            dt if dt.is_numeric() => apply_method_physical_integer!(s, agg_mean, groups),
+            #[cfg(feature = "dtype-datetime")]
+            dt @ Datetime(_, _) => self
+                .to_physical_repr()
+                .agg_mean(groups)
+                .cast(&Int64)
+                .unwrap()
+                .cast(dt)
+                .unwrap(),
+            #[cfg(feature = "dtype-duration")]
+            dt @ Duration(_) => self
+                .to_physical_repr()
+                .agg_mean(groups)
+                .cast(&Int64)
+                .unwrap()
+                .cast(dt)
+                .unwrap(),
+            #[cfg(feature = "dtype-time")]
+            Time => self
+                .to_physical_repr()
+                .agg_mean(groups)
+                .cast(&Int64)
+                .unwrap()
+                .cast(&Time)
+                .unwrap(),
+            #[cfg(feature = "dtype-date")]
+            Date => (self
+                .to_physical_repr()
+                .agg_mean(groups)
+                .cast(&Float64)
+                .unwrap()
+                * (MS_IN_DAY as f64))
+                .cast(&Datetime(TimeUnit::Milliseconds, None))
+                .unwrap(),
+            _ => Series::full_null("", groups.len(), s.dtype()),
+        }
+    }
+
     #[doc(hidden)]
     pub unsafe fn agg_median(&self, groups: &GroupsProxy) -> Series {
         // Prevent a rechunk for every individual group.
@@ -143,21 +195,38 @@ impl Series {
             Float64 => SeriesWrap(s.f64().unwrap().clone()).agg_median(groups),
             dt if dt.is_numeric() => apply_method_physical_integer!(s, agg_median, groups),
             #[cfg(feature = "dtype-datetime")]
-            dt @ (Datetime(_, _) | Duration(_) | Time) => s
+            dt @ Datetime(_, _) => self
                 .to_physical_repr()
                 .agg_median(groups)
                 .cast(&Int64)
                 .unwrap()
                 .cast(dt)
                 .unwrap(),
-            dt @ Date => {
-                let ca = s.to_physical_repr();
-                let physical_type = ca.dtype();
-                let s = apply_method_physical_integer!(ca, agg_median, groups);
-                // back to physical and then
-                // back to logical type
-                s.cast(physical_type).unwrap().cast(dt).unwrap()
-            },
+            #[cfg(feature = "dtype-duration")]
+            dt @ Duration(_) => self
+                .to_physical_repr()
+                .agg_median(groups)
+                .cast(&Int64)
+                .unwrap()
+                .cast(dt)
+                .unwrap(),
+            #[cfg(feature = "dtype-time")]
+            Time => self
+                .to_physical_repr()
+                .agg_median(groups)
+                .cast(&Int64)
+                .unwrap()
+                .cast(&Time)
+                .unwrap(),
+            #[cfg(feature = "dtype-date")]
+            Date => (self
+                .to_physical_repr()
+                .agg_median(groups)
+                .cast(&Float64)
+                .unwrap()
+                * (MS_IN_DAY as f64))
+                .cast(&Datetime(TimeUnit::Milliseconds, None))
+                .unwrap(),
             _ => Series::full_null("", groups.len(), s.dtype()),
         }
     }
@@ -197,41 +266,6 @@ impl Series {
         }
     }
 
-    #[doc(hidden)]
-    pub unsafe fn agg_mean(&self, groups: &GroupsProxy) -> Series {
-        // Prevent a rechunk for every individual group.
-        let s = if groups.len() > 1 {
-            self.rechunk()
-        } else {
-            self.clone()
-        };
-
-        use DataType::*;
-        match s.dtype() {
-            Boolean => s.cast(&Float64).unwrap().agg_mean(groups),
-            Float32 => SeriesWrap(s.f32().unwrap().clone()).agg_mean(groups),
-            Float64 => SeriesWrap(s.f64().unwrap().clone()).agg_mean(groups),
-            dt if dt.is_numeric() => apply_method_physical_integer!(s, agg_mean, groups),
-            #[cfg(feature = "dtype-datetime")]
-            dt @ (Datetime(_, _) | Duration(_) | Time) => s
-                .to_physical_repr()
-                .agg_mean(groups)
-                .cast(&Int64)
-                .unwrap()
-                .cast(dt)
-                .unwrap(),
-            dt @ Date => {
-                let ca = s.to_physical_repr();
-                let physical_type = ca.dtype();
-                let s = apply_method_physical_integer!(ca, agg_mean, groups);
-                // back to physical and then
-                // back to logical type
-                s.cast(physical_type).unwrap().cast(dt).unwrap()
-            },
-            _ => Series::full_null("", groups.len(), s.dtype()),
-        }
-    }
-
     #[doc(hidden)]
     pub unsafe fn agg_last(&self, groups: &GroupsProxy) -> Series {
         // Prevent a rechunk for every individual group.
diff --git a/crates/polars-core/src/series/implementations/date.rs b/crates/polars-core/src/series/implementations/date.rs
index 2b175ade5021..bb0058431120 100644
--- a/crates/polars-core/src/series/implementations/date.rs
+++ b/crates/polars-core/src/series/implementations/date.rs
@@ -317,11 +317,14 @@ impl SeriesTrait for SeriesWrap<DateChunked> {
     }
 
     fn median_reduce(&self) -> PolarsResult<Scalar> {
-        let av = AnyValue::from(self.median().map(|v| v as i64))
-            .cast(self.dtype())
-            .into_static()
-            .unwrap();
-        Ok(Scalar::new(self.dtype().clone(), av))
+        let av: AnyValue = self
+            .median()
+            .map(|v| (v * (MS_IN_DAY as f64)) as i64)
+            .into();
+        Ok(Scalar::new(
+            DataType::Datetime(TimeUnit::Milliseconds, None),
+            av,
+        ))
     }
 
     fn clone_inner(&self) -> Arc<dyn SeriesTrait> {
diff --git a/crates/polars-core/src/series/mod.rs b/crates/polars-core/src/series/mod.rs
index e6b392686d69..177b496f55cf 100644
--- a/crates/polars-core/src/series/mod.rs
+++ b/crates/polars-core/src/series/mod.rs
@@ -889,7 +889,26 @@ impl Series {
                 let val = self.mean();
                 Scalar::new(DataType::Float64, val.into())
             },
-            dt if dt.is_temporal() => {
+            #[cfg(feature = "dtype-date")]
+            DataType::Date => {
+                let val = self.mean().map(|v| (v * MS_IN_DAY as f64) as i64);
+                let av: AnyValue = val.into();
+                Scalar::new(DataType::Datetime(TimeUnit::Milliseconds, None), av)
+            },
+            #[cfg(feature = "dtype-datetime")]
+            dt @ DataType::Datetime(_, _) => {
+                let val = self.mean().map(|v| v as i64);
+                let av: AnyValue = val.into();
+                Scalar::new(dt.clone(), av)
+            },
+            #[cfg(feature = "dtype-duration")]
+            dt @ DataType::Duration(_) => {
+                let val = self.mean().map(|v| v as i64);
+                let av: AnyValue = val.into();
+                Scalar::new(dt.clone(), av)
+            },
+            #[cfg(feature = "dtype-time")]
+            dt @ DataType::Time => {
                 let val = self.mean().map(|v| v as i64);
                 let av: AnyValue = val.into();
                 Scalar::new(dt.clone(), av)
diff --git a/crates/polars-pipe/src/executors/sinks/group_by/aggregates/convert.rs b/crates/polars-pipe/src/executors/sinks/group_by/aggregates/convert.rs
index e928e2ba8a08..c3bc052d4b2a 100644
--- a/crates/polars-pipe/src/executors/sinks/group_by/aggregates/convert.rs
+++ b/crates/polars-pipe/src/executors/sinks/group_by/aggregates/convert.rs
@@ -241,7 +241,7 @@ where
                 #[cfg(feature = "dtype-categorical")]
                 if matches!(
                     logical_dtype,
-                    DataType::Categorical(_, _) | DataType::Enum(_, _)
+                    DataType::Categorical(_, _) | DataType::Enum(_, _) | DataType::Date
                 ) {
                     return (
                         logical_dtype.clone(),
diff --git a/crates/polars-plan/src/logical_plan/aexpr/schema.rs b/crates/polars-plan/src/logical_plan/aexpr/schema.rs
index c339fac4ad9e..8abe59d363ab 100644
--- a/crates/polars-plan/src/logical_plan/aexpr/schema.rs
+++ b/crates/polars-plan/src/logical_plan/aexpr/schema.rs
@@ -135,13 +135,19 @@ impl AExpr {
                     Median(expr) => {
                         let mut field =
                             arena.get(*expr).to_field(schema, Context::Default, arena)?;
-                        float_type(&mut field);
+                        match field.dtype {
+                            Date => field.coerce(Datetime(TimeUnit::Milliseconds, None)),
+                            _ => float_type(&mut field),
+                        }
                         Ok(field)
                     },
                     Mean(expr) => {
                         let mut field =
                             arena.get(*expr).to_field(schema, Context::Default, arena)?;
-                        float_type(&mut field);
+                        match field.dtype {
+                            Date => field.coerce(Datetime(TimeUnit::Milliseconds, None)),
+                            _ => float_type(&mut field),
+                        }
                         Ok(field)
                     },
                     Implode(expr) => {
diff --git a/crates/polars-plan/src/logical_plan/conversion/dsl_to_ir.rs b/crates/polars-plan/src/logical_plan/conversion/dsl_to_ir.rs
index 33be0ded29ab..f5f7bb959e91 100644
--- a/crates/polars-plan/src/logical_plan/conversion/dsl_to_ir.rs
+++ b/crates/polars-plan/src/logical_plan/conversion/dsl_to_ir.rs
@@ -471,16 +471,7 @@ pub fn to_alp_impl(
                             &input_schema,
                         ),
                         StatsFunction::Mean => stats_helper(
-                            |dt| {
-                                dt.is_numeric()
-                                    || matches!(
-                                        dt,
-                                        DataType::Boolean
-                                            | DataType::Duration(_)
-                                            | DataType::Datetime(_, _)
-                                            | DataType::Time
-                                    )
-                            },
+                            |dt| dt.is_numeric() || dt.is_temporal() || dt == &DataType::Boolean,
                             |name| col(name).mean(),
                             &input_schema,
                         ),
@@ -500,16 +491,7 @@ pub fn to_alp_impl(
                             stats_helper(|dt| dt.is_ord(), |name| col(name).max(), &input_schema)
                         },
                         StatsFunction::Median => stats_helper(
-                            |dt| {
-                                dt.is_numeric()
-                                    || matches!(
-                                        dt,
-                                        DataType::Boolean
-                                            | DataType::Duration(_)
-                                            | DataType::Datetime(_, _)
-                                            | DataType::Time
-                                    )
-                            },
+                            |dt| dt.is_numeric() || dt.is_temporal() || dt == &DataType::Boolean,
                             |name| col(name).median(),
                             &input_schema,
                         ),
diff --git a/py-polars/polars/dataframe/frame.py b/py-polars/polars/dataframe/frame.py
index ed073cfaa4ed..67b5e9cab6d5 100644
--- a/py-polars/polars/dataframe/frame.py
+++ b/py-polars/polars/dataframe/frame.py
@@ -4411,21 +4411,21 @@ def describe(
 
         >>> df.describe()
         shape: (9, 7)
-        ┌────────────┬──────────┬──────────┬──────────┬──────┬────────────┬──────────┐
-        │ statistic  ┆ float    ┆ int      ┆ bool     ┆ str  ┆ date       ┆ time     │
-        │ ---        ┆ ---      ┆ ---      ┆ ---      ┆ ---  ┆ ---        ┆ ---      │
-        │ str        ┆ f64      ┆ f64      ┆ f64      ┆ str  ┆ str        ┆ str      │
-        ╞════════════╪══════════╪══════════╪══════════╪══════╪════════════╪══════════╡
-        │ count      ┆ 3.0      ┆ 2.0      ┆ 3.0      ┆ 3    ┆ 3          ┆ 3        │
-        │ null_count ┆ 0.0      ┆ 1.0      ┆ 0.0      ┆ 0    ┆ 0          ┆ 0        │
-        │ mean       ┆ 2.266667 ┆ 45.0     ┆ 0.666667 ┆ null ┆ 2021-07-02 ┆ 16:07:10 │
-        │ std        ┆ 1.101514 ┆ 7.071068 ┆ null     ┆ null ┆ null       ┆ null     │
-        │ min        ┆ 1.0      ┆ 40.0     ┆ 0.0      ┆ xx   ┆ 2020-01-01 ┆ 10:20:30 │
-        │ 25%        ┆ 2.8      ┆ 40.0     ┆ null     ┆ null ┆ 2021-07-05 ┆ 14:45:50 │
-        │ 50%        ┆ 2.8      ┆ 50.0     ┆ null     ┆ null ┆ 2021-07-05 ┆ 14:45:50 │
-        │ 75%        ┆ 3.0      ┆ 50.0     ┆ null     ┆ null ┆ 2022-12-31 ┆ 23:15:10 │
-        │ max        ┆ 3.0      ┆ 50.0     ┆ 1.0      ┆ zz   ┆ 2022-12-31 ┆ 23:15:10 │
-        └────────────┴──────────┴──────────┴──────────┴──────┴────────────┴──────────┘
+        ┌────────────┬──────────┬──────────┬──────────┬──────┬─────────────────────┬──────────┐
+        │ statistic  ┆ float    ┆ int      ┆ bool     ┆ str  ┆ date                ┆ time     │
+        │ ---        ┆ ---      ┆ ---      ┆ ---      ┆ ---  ┆ ---                 ┆ ---      │
+        │ str        ┆ f64      ┆ f64      ┆ f64      ┆ str  ┆ str                 ┆ str      │
+        ╞════════════╪══════════╪══════════╪══════════╪══════╪═════════════════════╪══════════╡
+        │ count      ┆ 3.0      ┆ 2.0      ┆ 3.0      ┆ 3    ┆ 3                   ┆ 3        │
+        │ null_count ┆ 0.0      ┆ 1.0      ┆ 0.0      ┆ 0    ┆ 0                   ┆ 0        │
+        │ mean       ┆ 2.266667 ┆ 45.0     ┆ 0.666667 ┆ null ┆ 2021-07-02 16:00:00 ┆ 16:07:10 │
+        │ std        ┆ 1.101514 ┆ 7.071068 ┆ null     ┆ null ┆ null                ┆ null     │
+        │ min        ┆ 1.0      ┆ 40.0     ┆ 0.0      ┆ xx   ┆ 2020-01-01          ┆ 10:20:30 │
+        │ 25%        ┆ 2.8      ┆ 40.0     ┆ null     ┆ null ┆ 2021-07-05          ┆ 14:45:50 │
+        │ 50%        ┆ 2.8      ┆ 50.0     ┆ null     ┆ null ┆ 2021-07-05          ┆ 14:45:50 │
+        │ 75%        ┆ 3.0      ┆ 50.0     ┆ null     ┆ null ┆ 2022-12-31          ┆ 23:15:10 │
+        │ max        ┆ 3.0      ┆ 50.0     ┆ 1.0      ┆ zz   ┆ 2022-12-31          ┆ 23:15:10 │
+        └────────────┴──────────┴──────────┴──────────┴──────┴─────────────────────┴──────────┘
 
         Customize which percentiles are displayed, applying linear interpolation:
 
@@ -4435,24 +4435,24 @@ def describe(
         ...         interpolation="linear",
         ...     )
         shape: (11, 7)
-        ┌────────────┬──────────┬──────────┬──────────┬──────┬────────────┬──────────┐
-        │ statistic  ┆ float    ┆ int      ┆ bool     ┆ str  ┆ date       ┆ time     │
-        │ ---        ┆ ---      ┆ ---      ┆ ---      ┆ ---  ┆ ---        ┆ ---      │
-        │ str        ┆ f64      ┆ f64      ┆ f64      ┆ str  ┆ str        ┆ str      │
-        ╞════════════╪══════════╪══════════╪══════════╪══════╪════════════╪══════════╡
-        │ count      ┆ 3.0      ┆ 2.0      ┆ 3.0      ┆ 3    ┆ 3          ┆ 3        │
-        │ null_count ┆ 0.0      ┆ 1.0      ┆ 0.0      ┆ 0    ┆ 0          ┆ 0        │
-        │ mean       ┆ 2.266667 ┆ 45.0     ┆ 0.666667 ┆ null ┆ 2021-07-02 ┆ 16:07:10 │
-        │ std        ┆ 1.101514 ┆ 7.071068 ┆ null     ┆ null ┆ null       ┆ null     │
-        │ min        ┆ 1.0      ┆ 40.0     ┆ 0.0      ┆ xx   ┆ 2020-01-01 ┆ 10:20:30 │
-        │ 10%        ┆ 1.36     ┆ 41.0     ┆ null     ┆ null ┆ 2020-04-20 ┆ 11:13:34 │
-        │ 30%        ┆ 2.08     ┆ 43.0     ┆ null     ┆ null ┆ 2020-11-26 ┆ 12:59:42 │
-        │ 50%        ┆ 2.8      ┆ 45.0     ┆ null     ┆ null ┆ 2021-07-05 ┆ 14:45:50 │
-        │ 70%        ┆ 2.88     ┆ 47.0     ┆ null     ┆ null ┆ 2022-02-07 ┆ 18:09:34 │
-        │ 90%        ┆ 2.96     ┆ 49.0     ┆ null     ┆ null ┆ 2022-09-13 ┆ 21:33:18 │
-        │ max        ┆ 3.0      ┆ 50.0     ┆ 1.0      ┆ zz   ┆ 2022-12-31 ┆ 23:15:10 │
-        └────────────┴──────────┴──────────┴──────────┴──────┴────────────┴──────────┘
-        """
+        ┌────────────┬──────────┬──────────┬──────────┬──────┬─────────────────────┬──────────┐
+        │ statistic  ┆ float    ┆ int      ┆ bool     ┆ str  ┆ date                ┆ time     │
+        │ ---        ┆ ---      ┆ ---      ┆ ---      ┆ ---  ┆ ---                 ┆ ---      │
+        │ str        ┆ f64      ┆ f64      ┆ f64      ┆ str  ┆ str                 ┆ str      │
+        ╞════════════╪══════════╪══════════╪══════════╪══════╪═════════════════════╪══════════╡
+        │ count      ┆ 3.0      ┆ 2.0      ┆ 3.0      ┆ 3    ┆ 3                   ┆ 3        │
+        │ null_count ┆ 0.0      ┆ 1.0      ┆ 0.0      ┆ 0    ┆ 0                   ┆ 0        │
+        │ mean       ┆ 2.266667 ┆ 45.0     ┆ 0.666667 ┆ null ┆ 2021-07-02 16:00:00 ┆ 16:07:10 │
+        │ std        ┆ 1.101514 ┆ 7.071068 ┆ null     ┆ null ┆ null                ┆ null     │
+        │ min        ┆ 1.0      ┆ 40.0     ┆ 0.0      ┆ xx   ┆ 2020-01-01          ┆ 10:20:30 │
+        │ 10%        ┆ 1.36     ┆ 41.0     ┆ null     ┆ null ┆ 2020-04-20          ┆ 11:13:34 │
+        │ 30%        ┆ 2.08     ┆ 43.0     ┆ null     ┆ null ┆ 2020-11-26          ┆ 12:59:42 │
+        │ 50%        ┆ 2.8      ┆ 45.0     ┆ null     ┆ null ┆ 2021-07-05          ┆ 14:45:50 │
+        │ 70%        ┆ 2.88     ┆ 47.0     ┆ null     ┆ null ┆ 2022-02-07          ┆ 18:09:34 │
+        │ 90%        ┆ 2.96     ┆ 49.0     ┆ null     ┆ null ┆ 2022-09-13          ┆ 21:33:18 │
+        │ max        ┆ 3.0      ┆ 50.0     ┆ 1.0      ┆ zz   ┆ 2022-12-31          ┆ 23:15:10 │
+        └────────────┴──────────┴──────────┴──────────┴──────┴─────────────────────┴──────────┘
+        """  # noqa: W505
         if not self.columns:
             msg = "cannot describe a DataFrame that has no columns"
             raise TypeError(msg)
diff --git a/py-polars/polars/lazyframe/frame.py b/py-polars/polars/lazyframe/frame.py
index ba198093c4df..01acd9b6514c 100644
--- a/py-polars/polars/lazyframe/frame.py
+++ b/py-polars/polars/lazyframe/frame.py
@@ -772,21 +772,21 @@ def describe(
 
         >>> lf.describe()
         shape: (9, 7)
-        ┌────────────┬──────────┬──────────┬──────────┬──────┬────────────┬──────────┐
-        │ statistic  ┆ float    ┆ int      ┆ bool     ┆ str  ┆ date       ┆ time     │
-        │ ---        ┆ ---      ┆ ---      ┆ ---      ┆ ---  ┆ ---        ┆ ---      │
-        │ str        ┆ f64      ┆ f64      ┆ f64      ┆ str  ┆ str        ┆ str      │
-        ╞════════════╪══════════╪══════════╪══════════╪══════╪════════════╪══════════╡
-        │ count      ┆ 3.0      ┆ 2.0      ┆ 3.0      ┆ 3    ┆ 3          ┆ 3        │
-        │ null_count ┆ 0.0      ┆ 1.0      ┆ 0.0      ┆ 0    ┆ 0          ┆ 0        │
-        │ mean       ┆ 2.266667 ┆ 45.0     ┆ 0.666667 ┆ null ┆ 2021-07-02 ┆ 16:07:10 │
-        │ std        ┆ 1.101514 ┆ 7.071068 ┆ null     ┆ null ┆ null       ┆ null     │
-        │ min        ┆ 1.0      ┆ 40.0     ┆ 0.0      ┆ xx   ┆ 2020-01-01 ┆ 10:20:30 │
-        │ 25%        ┆ 2.8      ┆ 40.0     ┆ null     ┆ null ┆ 2021-07-05 ┆ 14:45:50 │
-        │ 50%        ┆ 2.8      ┆ 50.0     ┆ null     ┆ null ┆ 2021-07-05 ┆ 14:45:50 │
-        │ 75%        ┆ 3.0      ┆ 50.0     ┆ null     ┆ null ┆ 2022-12-31 ┆ 23:15:10 │
-        │ max        ┆ 3.0      ┆ 50.0     ┆ 1.0      ┆ zz   ┆ 2022-12-31 ┆ 23:15:10 │
-        └────────────┴──────────┴──────────┴──────────┴──────┴────────────┴──────────┘
+        ┌────────────┬──────────┬──────────┬──────────┬──────┬─────────────────────┬──────────┐
+        │ statistic  ┆ float    ┆ int      ┆ bool     ┆ str  ┆ date                ┆ time     │
+        │ ---        ┆ ---      ┆ ---      ┆ ---      ┆ ---  ┆ ---                 ┆ ---      │
+        │ str        ┆ f64      ┆ f64      ┆ f64      ┆ str  ┆ str                 ┆ str      │
+        ╞════════════╪══════════╪══════════╪══════════╪══════╪═════════════════════╪══════════╡
+        │ count      ┆ 3.0      ┆ 2.0      ┆ 3.0      ┆ 3    ┆ 3                   ┆ 3        │
+        │ null_count ┆ 0.0      ┆ 1.0      ┆ 0.0      ┆ 0    ┆ 0                   ┆ 0        │
+        │ mean       ┆ 2.266667 ┆ 45.0     ┆ 0.666667 ┆ null ┆ 2021-07-02 16:00:00 ┆ 16:07:10 │
+        │ std        ┆ 1.101514 ┆ 7.071068 ┆ null     ┆ null ┆ null                ┆ null     │
+        │ min        ┆ 1.0      ┆ 40.0     ┆ 0.0      ┆ xx   ┆ 2020-01-01          ┆ 10:20:30 │
+        │ 25%        ┆ 2.8      ┆ 40.0     ┆ null     ┆ null ┆ 2021-07-05          ┆ 14:45:50 │
+        │ 50%        ┆ 2.8      ┆ 50.0     ┆ null     ┆ null ┆ 2021-07-05          ┆ 14:45:50 │
+        │ 75%        ┆ 3.0      ┆ 50.0     ┆ null     ┆ null ┆ 2022-12-31          ┆ 23:15:10 │
+        │ max        ┆ 3.0      ┆ 50.0     ┆ 1.0      ┆ zz   ┆ 2022-12-31          ┆ 23:15:10 │
+        └────────────┴──────────┴──────────┴──────────┴──────┴─────────────────────┴──────────┘
 
         Customize which percentiles are displayed, applying linear interpolation:
 
@@ -796,24 +796,24 @@ def describe(
         ...         interpolation="linear",
         ...     )
         shape: (11, 7)
-        ┌────────────┬──────────┬──────────┬──────────┬──────┬────────────┬──────────┐
-        │ statistic  ┆ float    ┆ int      ┆ bool     ┆ str  ┆ date       ┆ time     │
-        │ ---        ┆ ---      ┆ ---      ┆ ---      ┆ ---  ┆ ---        ┆ ---      │
-        │ str        ┆ f64      ┆ f64      ┆ f64      ┆ str  ┆ str        ┆ str      │
-        ╞════════════╪══════════╪══════════╪══════════╪══════╪════════════╪══════════╡
-        │ count      ┆ 3.0      ┆ 2.0      ┆ 3.0      ┆ 3    ┆ 3          ┆ 3        │
-        │ null_count ┆ 0.0      ┆ 1.0      ┆ 0.0      ┆ 0    ┆ 0          ┆ 0        │
-        │ mean       ┆ 2.266667 ┆ 45.0     ┆ 0.666667 ┆ null ┆ 2021-07-02 ┆ 16:07:10 │
-        │ std        ┆ 1.101514 ┆ 7.071068 ┆ null     ┆ null ┆ null       ┆ null     │
-        │ min        ┆ 1.0      ┆ 40.0     ┆ 0.0      ┆ xx   ┆ 2020-01-01 ┆ 10:20:30 │
-        │ 10%        ┆ 1.36     ┆ 41.0     ┆ null     ┆ null ┆ 2020-04-20 ┆ 11:13:34 │
-        │ 30%        ┆ 2.08     ┆ 43.0     ┆ null     ┆ null ┆ 2020-11-26 ┆ 12:59:42 │
-        │ 50%        ┆ 2.8      ┆ 45.0     ┆ null     ┆ null ┆ 2021-07-05 ┆ 14:45:50 │
-        │ 70%        ┆ 2.88     ┆ 47.0     ┆ null     ┆ null ┆ 2022-02-07 ┆ 18:09:34 │
-        │ 90%        ┆ 2.96     ┆ 49.0     ┆ null     ┆ null ┆ 2022-09-13 ┆ 21:33:18 │
-        │ max        ┆ 3.0      ┆ 50.0     ┆ 1.0      ┆ zz   ┆ 2022-12-31 ┆ 23:15:10 │
-        └────────────┴──────────┴──────────┴──────────┴──────┴────────────┴──────────┘
-        """
+        ┌────────────┬──────────┬──────────┬──────────┬──────┬─────────────────────┬──────────┐
+        │ statistic  ┆ float    ┆ int      ┆ bool     ┆ str  ┆ date                ┆ time     │
+        │ ---        ┆ ---      ┆ ---      ┆ ---      ┆ ---  ┆ ---                 ┆ ---      │
+        │ str        ┆ f64      ┆ f64      ┆ f64      ┆ str  ┆ str                 ┆ str      │
+        ╞════════════╪══════════╪══════════╪══════════╪══════╪═════════════════════╪══════════╡
+        │ count      ┆ 3.0      ┆ 2.0      ┆ 3.0      ┆ 3    ┆ 3                   ┆ 3        │
+        │ null_count ┆ 0.0      ┆ 1.0      ┆ 0.0      ┆ 0    ┆ 0                   ┆ 0        │
+        │ mean       ┆ 2.266667 ┆ 45.0     ┆ 0.666667 ┆ null ┆ 2021-07-02 16:00:00 ┆ 16:07:10 │
+        │ std        ┆ 1.101514 ┆ 7.071068 ┆ null     ┆ null ┆ null                ┆ null     │
+        │ min        ┆ 1.0      ┆ 40.0     ┆ 0.0      ┆ xx   ┆ 2020-01-01          ┆ 10:20:30 │
+        │ 10%        ┆ 1.36     ┆ 41.0     ┆ null     ┆ null ┆ 2020-04-20          ┆ 11:13:34 │
+        │ 30%        ┆ 2.08     ┆ 43.0     ┆ null     ┆ null ┆ 2020-11-26          ┆ 12:59:42 │
+        │ 50%        ┆ 2.8      ┆ 45.0     ┆ null     ┆ null ┆ 2021-07-05          ┆ 14:45:50 │
+        │ 70%        ┆ 2.88     ┆ 47.0     ┆ null     ┆ null ┆ 2022-02-07          ┆ 18:09:34 │
+        │ 90%        ┆ 2.96     ┆ 49.0     ┆ null     ┆ null ┆ 2022-09-13          ┆ 21:33:18 │
+        │ max        ┆ 3.0      ┆ 50.0     ┆ 1.0      ┆ zz   ┆ 2022-12-31          ┆ 23:15:10 │
+        └────────────┴──────────┴──────────┴──────────┴──────┴─────────────────────┴──────────┘
+        """  # noqa: W505
         from polars.convert import from_dict
 
         if not self.columns:
@@ -847,9 +847,9 @@ def skip_minmax(dt: PolarsDataType) -> bool:
             ]
             # mean
             mean_expr = (
-                F.col(c).to_physical().mean().cast(dtype)
-                if is_temporal
-                else (F.col(c).mean() if is_numeric or dtype == Boolean else null)
+                F.col(c).mean()
+                if is_temporal or is_numeric or dtype == Boolean
+                else null
             )
 
             # standard deviation, min, max
@@ -910,9 +910,11 @@ def skip_minmax(dt: PolarsDataType) -> bool:
         # cast by column type (numeric/bool -> float), (other -> string)
         for c in self.columns:
             summary[c] = [  # type: ignore[assignment]
-                None
-                if (v is None or isinstance(v, dict))
-                else (float(v) if (c in has_numeric_result) else str(v))
+                (
+                    None
+                    if (v is None or isinstance(v, dict))
+                    else (float(v) if (c in has_numeric_result) else str(v))
+                )
                 for v in summary[c]
             ]
 
diff --git a/py-polars/polars/series/datetime.py b/py-polars/polars/series/datetime.py
index 8a1378d06451..059cea931511 100644
--- a/py-polars/polars/series/datetime.py
+++ b/py-polars/polars/series/datetime.py
@@ -2,11 +2,9 @@
 
 from typing import TYPE_CHECKING, Iterable
 
-from polars._utils.convert import to_py_date, to_py_datetime
 from polars._utils.deprecation import deprecate_function
 from polars._utils.unstable import unstable
 from polars._utils.wrap import wrap_s
-from polars.datatypes import Date, Datetime, Duration, Time
 from polars.series.utils import expr_dispatch
 
 if TYPE_CHECKING:
@@ -156,13 +154,16 @@ def max(self) -> dt.date | dt.datetime | dt.timedelta | None:
         """
         return wrap_s(self._s).max()  # type: ignore[return-value]
 
-    def median(self) -> TemporalLiteral | float | None:
+    def median(self) -> TemporalLiteral | None:
         """
         Return median as python DateTime.
 
         Examples
         --------
-        >>> from datetime import datetime
+        >>> from datetime import date, datetime
+        >>> s = pl.Series([date(2001, 1, 1), date(2001, 1, 2)])
+        >>> s.dt.median()
+        datetime.datetime(2001, 1, 1, 12, 0)
         >>> date = pl.datetime_range(
         ...     datetime(2001, 1, 1), datetime(2001, 1, 3), "1d", eager=True
         ... ).alias("datetime")
@@ -177,40 +178,25 @@ def median(self) -> TemporalLiteral | float | None:
         >>> date.dt.median()
         datetime.datetime(2001, 1, 2, 0, 0)
         """
-        s = wrap_s(self._s)
-        out = s.median()
-        if out is not None:
-            if s.dtype == Date:
-                return to_py_date(int(out))  # type: ignore[arg-type]
-            elif s.dtype in (Datetime, Duration, Time):
-                return out  # type: ignore[return-value]
-            else:
-                return to_py_datetime(int(out), s.dtype.time_unit)  # type: ignore[arg-type, attr-defined]
-        return None
-
-    def mean(self) -> TemporalLiteral | float | None:
+        return self._s.median()
+
+    def mean(self) -> TemporalLiteral | None:
         """
         Return mean as python DateTime.
 
         Examples
         --------
-        >>> from datetime import datetime
+        >>> from datetime import date, datetime
+        >>> s = pl.Series([date(2001, 1, 1), date(2001, 1, 2)])
+        >>> s.dt.mean()
+        datetime.datetime(2001, 1, 1, 12, 0)
         >>> s = pl.Series(
         ...     [datetime(2001, 1, 1), datetime(2001, 1, 2), datetime(2001, 1, 3)]
         ... )
         >>> s.dt.mean()
         datetime.datetime(2001, 1, 2, 0, 0)
         """
-        s = wrap_s(self._s)
-        out = s.mean()
-        if out is not None:
-            if s.dtype == Date:
-                return to_py_date(int(out))  # type: ignore[arg-type]
-            elif s.dtype in (Datetime, Duration, Time):
-                return out  # type: ignore[return-value]
-            else:
-                return to_py_datetime(int(out), s.dtype.time_unit)  # type: ignore[arg-type, attr-defined]
-        return None
+        return self._s.mean()
 
     def to_string(self, format: str) -> Series:
         """
diff --git a/py-polars/src/series/aggregation.rs b/py-polars/src/series/aggregation.rs
index 5874ec08ac38..d8a5364afe3b 100644
--- a/py-polars/src/series/aggregation.rs
+++ b/py-polars/src/series/aggregation.rs
@@ -1,4 +1,5 @@
 use pyo3::prelude::*;
+use DataType::*;
 
 use crate::error::PyPolarsErr;
 use crate::prelude::*;
@@ -44,7 +45,7 @@ impl PySeries {
 
     fn mean(&self, py: Python) -> PyResult<PyObject> {
         match self.series.dtype() {
-            DataType::Boolean => Ok(Wrap(
+            Boolean => Ok(Wrap(
                 self.series
                     .cast(&DataType::UInt8)
                     .unwrap()
@@ -52,7 +53,8 @@ impl PySeries {
                     .as_any_value(),
             )
             .into_py(py)),
-            DataType::Datetime(_, _) | DataType::Duration(_) | DataType::Time => {
+            // For non-numeric output types we require mean_reduce.
+            dt if dt.is_temporal() => {
                 Ok(Wrap(self.series.mean_reduce().as_any_value()).into_py(py))
             },
             _ => Ok(self.series.mean().into_py(py)),
@@ -61,7 +63,7 @@ impl PySeries {
 
     fn median(&self, py: Python) -> PyResult<PyObject> {
         match self.series.dtype() {
-            DataType::Boolean => Ok(Wrap(
+            Boolean => Ok(Wrap(
                 self.series
                     .cast(&DataType::UInt8)
                     .unwrap()
@@ -70,7 +72,8 @@ impl PySeries {
                     .as_any_value(),
             )
             .into_py(py)),
-            DataType::Datetime(_, _) | DataType::Duration(_) | DataType::Time => Ok(Wrap(
+            // For non-numeric output types we require median_reduce.
+            dt if dt.is_temporal() => Ok(Wrap(
                 self.series
                     .median_reduce()
                     .map_err(PyPolarsErr::from)?
diff --git a/py-polars/tests/unit/dataframe/test_describe.py b/py-polars/tests/unit/dataframe/test_describe.py
index 2c38aec302e9..edeee04ae3ed 100644
--- a/py-polars/tests/unit/dataframe/test_describe.py
+++ b/py-polars/tests/unit/dataframe/test_describe.py
@@ -75,7 +75,7 @@ def test_df_describe(lazy: bool) -> None:
             "g": [
                 "3",
                 "0",
-                "2021-07-02",
+                "2021-07-02 16:00:00",
                 None,
                 "2020-01-01",
                 "2021-07-05",
diff --git a/py-polars/tests/unit/operations/namespaces/temporal/test_datetime.py b/py-polars/tests/unit/operations/namespaces/temporal/test_datetime.py
index f2eed19af83e..3228f94ec5d9 100644
--- a/py-polars/tests/unit/operations/namespaces/temporal/test_datetime.py
+++ b/py-polars/tests/unit/operations/namespaces/temporal/test_datetime.py
@@ -1001,9 +1001,9 @@ def test_weekday(time_unit: TimeUnit) -> None:
     [
         ([], None),
         ([None, None], None),
-        ([date(2022, 1, 1)], date(2022, 1, 1)),
-        ([date(2022, 1, 1), date(2022, 1, 2), date(2022, 1, 3)], date(2022, 1, 2)),
-        ([date(2022, 1, 1), date(2022, 1, 2), date(2024, 5, 15)], date(2022, 1, 2)),
+        ([date(2022, 1, 1)], datetime(2022, 1, 1)),
+        ([date(2022, 1, 1), date(2022, 1, 2), date(2022, 1, 4)], datetime(2022, 1, 2)),
+        ([date(2022, 1, 1), date(2022, 1, 2), date(2024, 5, 15)], datetime(2022, 1, 2)),
         ([datetime(2022, 1, 1)], datetime(2022, 1, 1)),
         (
             [datetime(2022, 1, 1), datetime(2022, 1, 2), datetime(2022, 1, 3)],
@@ -1052,9 +1052,15 @@ def test_median(
     [
         ([], None),
         ([None, None], None),
-        ([date(2022, 1, 1)], date(2022, 1, 1)),
-        ([date(2022, 1, 1), date(2022, 1, 2), date(2022, 1, 3)], date(2022, 1, 2)),
-        ([date(2022, 1, 1), date(2022, 1, 2), date(2024, 5, 15)], date(2022, 10, 16)),
+        ([date(2022, 1, 1)], datetime(2022, 1, 1)),
+        (
+            [date(2022, 1, 1), date(2022, 1, 2), date(2022, 1, 4)],
+            datetime(2022, 1, 2, 8),
+        ),
+        (
+            [date(2022, 1, 1), date(2022, 1, 2), date(2024, 5, 15)],
+            datetime(2022, 10, 16, 16, 0),
+        ),
         ([datetime(2022, 1, 1)], datetime(2022, 1, 1)),
         (
             [datetime(2022, 1, 1), datetime(2022, 1, 2), datetime(2022, 1, 3)],
@@ -1181,6 +1187,10 @@ def test_duration_median_with_tu(
 def test_agg_mean_expr() -> None:
     df = pl.DataFrame(
         {
+            "date": pl.Series(
+                [date(2023, 1, 1), date(2023, 1, 2), date(2023, 1, 4)],
+                dtype=pl.Date,
+            ),
             "datetime_ms": pl.Series(
                 [datetime(2023, 1, 1), datetime(2023, 1, 2), datetime(2023, 1, 4)],
                 dtype=pl.Datetime("ms"),
@@ -1214,6 +1224,7 @@ def test_agg_mean_expr() -> None:
 
     expected = pl.DataFrame(
         {
+            "date": pl.Series([datetime(2023, 1, 2, 8, 0)], dtype=pl.Datetime("ms")),
             "datetime_ms": pl.Series(
                 [datetime(2023, 1, 2, 8, 0, 0)], dtype=pl.Datetime("ms")
             ),
@@ -1242,6 +1253,10 @@ def test_agg_mean_expr() -> None:
 def test_agg_median_expr() -> None:
     df = pl.DataFrame(
         {
+            "date": pl.Series(
+                [date(2023, 1, 1), date(2023, 1, 2), date(2023, 1, 4)],
+                dtype=pl.Date,
+            ),
             "datetime_ms": pl.Series(
                 [datetime(2023, 1, 1), datetime(2023, 1, 2), datetime(2023, 1, 4)],
                 dtype=pl.Datetime("ms"),
@@ -1275,6 +1290,7 @@ def test_agg_median_expr() -> None:
 
     expected = pl.DataFrame(
         {
+            "date": pl.Series([datetime(2023, 1, 2)], dtype=pl.Datetime("ms")),
             "datetime_ms": pl.Series([datetime(2023, 1, 2)], dtype=pl.Datetime("ms")),
             "datetime_us": pl.Series([datetime(2023, 1, 2)], dtype=pl.Datetime("us")),
             "datetime_ns": pl.Series([datetime(2023, 1, 2)], dtype=pl.Datetime("ns")),
diff --git a/py-polars/tests/unit/operations/test_group_by.py b/py-polars/tests/unit/operations/test_group_by.py
index 855d867889cb..e53fece337f7 100644
--- a/py-polars/tests/unit/operations/test_group_by.py
+++ b/py-polars/tests/unit/operations/test_group_by.py
@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from collections import OrderedDict
-from datetime import datetime, timedelta
+from datetime import date, datetime, timedelta
 from typing import TYPE_CHECKING, Any
 
 import numpy as np
@@ -67,6 +67,12 @@ def test_group_by() -> None:
         ([1, 2, 3, 4], [2, 4], pl.Float32, pl.Float32),
         ([1, 2, 3, 4], [2, 4], pl.Float64, pl.Float64),
         ([False, True, True, True], [2 / 3, 1], pl.Boolean, pl.Float64),
+        (
+            [date(2023, 1, 1), date(2023, 1, 2), date(2023, 1, 4), date(2023, 1, 5)],
+            [datetime(2023, 1, 2, 8, 0, 0), datetime(2023, 1, 5)],
+            pl.Date,
+            pl.Datetime("ms"),
+        ),
         (
             [
                 datetime(2023, 1, 1),
@@ -158,6 +164,12 @@ def test_group_by_mean_by_dtype(
         ([1, 2, 4, 5], [2, 5], pl.Float32, pl.Float32),
         ([1, 2, 4, 5], [2, 5], pl.Float64, pl.Float64),
         ([False, True, True, True], [1, 1], pl.Boolean, pl.Float64),
+        (
+            [date(2023, 1, 1), date(2023, 1, 2), date(2023, 1, 4), date(2023, 1, 5)],
+            [datetime(2023, 1, 2), datetime(2023, 1, 5)],
+            pl.Date,
+            pl.Datetime("ms"),
+        ),
         (
             [
                 datetime(2023, 1, 1),
diff --git a/py-polars/tests/unit/series/test_describe.py b/py-polars/tests/unit/series/test_describe.py
index cdf180423231..be0641a02fb7 100644
--- a/py-polars/tests/unit/series/test_describe.py
+++ b/py-polars/tests/unit/series/test_describe.py
@@ -81,7 +81,7 @@ def test_series_describe_date() -> None:
     stats = {
         "count": "3",
         "null_count": "0",
-        "mean": "2010-09-29",
+        "mean": "2010-09-29 16:00:00",
         "min": "1999-12-31",
         "25%": "2005-08-05",
         "50%": "2011-03-11",
diff --git a/py-polars/tests/unit/streaming/test_streaming_group_by.py b/py-polars/tests/unit/streaming/test_streaming_group_by.py
index 5e18040db398..2540cd7c0fd1 100644
--- a/py-polars/tests/unit/streaming/test_streaming_group_by.py
+++ b/py-polars/tests/unit/streaming/test_streaming_group_by.py
@@ -65,7 +65,8 @@ def test_streaming_group_by_types() -> None:
                         pl.col("bool").mean().alias("bool_mean"),
                         pl.col("bool").sum().alias("bool_sum"),
                         pl.col("date").sum().alias("date_sum"),
-                        pl.col("date").mean().alias("date_mean"),
+                        # Date streaming mean/median has been temporarily disabled
+                        # pl.col("date").mean().alias("date_mean"),
                         pl.col("date").first().alias("date_first"),
                         pl.col("date").last().alias("date_last"),
                         pl.col("date").min().alias("date_min"),
@@ -86,7 +87,7 @@ def test_streaming_group_by_types() -> None:
             "bool_mean": pl.Float64,
             "bool_sum": pl.UInt32,
             "date_sum": pl.Date,
-            "date_mean": pl.Date,
+            # "date_mean": pl.Date,
             "date_first": pl.Date,
             "date_last": pl.Date,
             "date_min": pl.Date,
@@ -103,7 +104,8 @@ def test_streaming_group_by_types() -> None:
             "bool_mean": [0.5],
             "bool_sum": [1],
             "date_sum": [date(2074, 1, 1)],
-            "date_mean": [date(2022, 1, 1)],
+            # Date streaming mean/median has been temporarily disabled
+            # "date_mean": [date(2022, 1, 1)],
             "date_first": [date(2022, 1, 1)],
             "date_last": [date(2022, 1, 1)],
             "date_min": [date(2022, 1, 1)],