diff --git a/py-polars/polars/_utils/construction/dataframe.py b/py-polars/polars/_utils/construction/dataframe.py
index 573d89890589..e5a30a4107d1 100644
--- a/py-polars/polars/_utils/construction/dataframe.py
+++ b/py-polars/polars/_utils/construction/dataframe.py
@@ -134,7 +134,7 @@ def dict_to_pydf(
     else:
         data_series = [
             s._s
-            for s in _expand_dict_scalars(
+            for s in _expand_dict_values(
                 data,
                 schema_overrides=schema_overrides,
                 strict=strict,
@@ -307,7 +307,7 @@ def _post_apply_columns(
     return pydf
 
 
-def _expand_dict_scalars(
+def _expand_dict_values(
     data: Mapping[str, Sequence[object] | Mapping[str, Sequence[object]] | Series],
     *,
     schema_overrides: SchemaDict | None = None,
@@ -334,9 +334,20 @@ def _expand_dict_scalars(
             for name, val in data.items():
                 dtype = dtypes.get(name)
                 if isinstance(val, dict) and dtype != Struct:
-                    updated_data[name] = pl.DataFrame(val, strict=strict).to_struct(
-                        name
-                    )
+                    vdf = pl.DataFrame(val, strict=strict)
+                    if (
+                        len(vdf) == 1
+                        and array_len > 1
+                        and all(not d.is_nested() for d in vdf.schema.values())
+                    ):
+                        s_vals = {
+                            nm: vdf[nm].extend_constant(v, n=(array_len - 1))
+                            for nm, v in val.items()
+                        }
+                        st = pl.DataFrame(s_vals).to_struct(name)
+                    else:
+                        st = vdf.to_struct(name)
+                    updated_data[name] = st
 
                 elif isinstance(val, pl.Series):
                     s = val.rename(name) if name != val.name else val
diff --git a/py-polars/polars/_utils/various.py b/py-polars/polars/_utils/various.py
index e689cf2ea632..db18f2adf7fc 100644
--- a/py-polars/polars/_utils/various.py
+++ b/py-polars/polars/_utils/various.py
@@ -207,9 +207,9 @@ def _in_notebook() -> bool:
 
 
 def arrlen(obj: Any) -> int | None:
-    """Return length of (non-string) sequence object; returns None for non-sequences."""
+    """Return length of (non-string/dict) sequence; returns None for non-sequences."""
     try:
-        return None if isinstance(obj, str) else len(obj)
+        return None if isinstance(obj, (str, dict)) else len(obj)
     except TypeError:
         return None
 
diff --git a/py-polars/tests/unit/dataframe/test_from_dict.py b/py-polars/tests/unit/dataframe/test_from_dict.py
index 012a27ca9ce5..effdc47a4d4d 100644
--- a/py-polars/tests/unit/dataframe/test_from_dict.py
+++ b/py-polars/tests/unit/dataframe/test_from_dict.py
@@ -146,7 +146,7 @@ def test_from_dict_with_scalars() -> None:
 
 
 @pytest.mark.slow()
-def test_from_dict_with_scalars_mixed() -> None:
+def test_from_dict_with_values_mixed() -> None:
     # a bit of everything
     mixed_dtype_data: dict[str, Any] = {
         "a": 0,
@@ -164,11 +164,10 @@ def test_from_dict_with_scalars_mixed() -> None:
     # note: deliberately set this value large; if all dtypes are
     # on the fast-path it'll only take ~0.03secs. if it becomes
     # even remotely noticeable that will indicate a regression.
-    # TODO: This is now slow (~0.15 seconds). Needs to be looked into.
     n_range = 1_000_000
     index_and_data: dict[str, Any] = {"idx": range(n_range)}
     index_and_data.update(mixed_dtype_data.items())
-    df8 = pl.DataFrame(
+    df = pl.DataFrame(
         data=index_and_data,
         schema={
             "idx": pl.Int32,
@@ -185,14 +184,46 @@ def test_from_dict_with_scalars_mixed() -> None:
             "k": pl.String,
         },
     )
-    dfx = df8.select(pl.exclude("idx"))
+    dfx = df.select(pl.exclude("idx"))
 
-    assert len(df8) == n_range
+    assert len(df) == n_range
     assert dfx[:5].rows() == dfx[5:10].rows()
     assert dfx[-10:-5].rows() == dfx[-5:].rows()
     assert dfx.row(n_range // 2, named=True) == mixed_dtype_data
 
 
+def test_from_dict_expand_nested_struct() -> None:
+    # confirm consistent init of nested struct from dict data
+    dt = date(2077, 10, 10)
+    expected = pl.DataFrame(
+        [
+            pl.Series("x", [dt]),
+            pl.Series("nested", [{"y": -1, "z": 1}]),
+        ]
+    )
+    for df in (
+        pl.DataFrame({"x": dt, "nested": {"y": -1, "z": 1}}),
+        pl.DataFrame({"x": dt, "nested": [{"y": -1, "z": 1}]}),
+        pl.DataFrame({"x": [dt], "nested": {"y": -1, "z": 1}}),
+        pl.DataFrame({"x": [dt], "nested": [{"y": -1, "z": 1}]}),
+    ):
+        assert_frame_equal(expected, df)
+
+    # confirm expansion to 'n' nested values
+    nested_values = [{"y": -1, "z": 1}, {"y": -1, "z": 1}, {"y": -1, "z": 1}]
+    expected = pl.DataFrame(
+        [
+            pl.Series("x", [0, 1, 2]),
+            pl.Series("nested", nested_values),
+        ]
+    )
+    for df in (
+        pl.DataFrame({"x": range(3), "nested": {"y": -1, "z": 1}}),
+        pl.DataFrame({"x": [0, 1, 2], "nested": {"y": -1, "z": 1}}),
+    ):
+        assert_frame_equal(expected, df)
+
+
 def test_from_dict_duration_subseconds() -> None:
     d = {"duration": [timedelta(seconds=1, microseconds=1000)]}
     result = pl.from_dict(d)