diff --git a/pyiceberg/io/pyarrow.py b/pyiceberg/io/pyarrow.py
index c36f1639d9..974dcf1023 100644
--- a/pyiceberg/io/pyarrow.py
+++ b/pyiceberg/io/pyarrow.py
@@ -2073,9 +2073,12 @@ def list(self, list_type: ListType, list_array: pa.Array | None, value_array: pa
         if isinstance(list_array, (pa.ListArray, pa.LargeListArray, pa.FixedSizeListArray)) and value_array is not None:
             list_initializer = pa.large_list if isinstance(list_array, pa.LargeListArray) else pa.list_
             if isinstance(value_array, pa.StructArray):
-                # This can be removed once this has been fixed:
-                # https://github.com/apache/arrow/issues/38809
-                list_array = pa.LargeListArray.from_arrays(list_array.offsets, value_array)
+                # Rebuilding is what applies the projection to the element struct. A plain
+                # cast would not do: Iceberg renames by field-id, while cast matches the
+                # element fields by name and would silently null out any renamed one. The
+                # mask has to be carried over because the offsets buffer alone cannot
+                # express a null list, only an empty one.
+                list_array = pa.LargeListArray.from_arrays(list_array.offsets, value_array, mask=list_array.is_null())
             value_array = self._cast_if_needed(list_type.element_field, value_array)
             arrow_field = list_initializer(self._construct_field(list_type.element_field, value_array.type))
             return list_array.cast(arrow_field)
diff --git a/tests/integration/test_reads.py b/tests/integration/test_reads.py
index 3ee94b751f..ef81add078 100644
--- a/tests/integration/test_reads.py
+++ b/tests/integration/test_reads.py
@@ -994,10 +994,7 @@ def test_null_list_and_map(catalog: Catalog) -> None:
     arrow_table = table_test_empty_list_and_map.scan().to_arrow()
     assert arrow_table["col_list"].to_pylist() == [None, []]
     assert arrow_table["col_map"].to_pylist() == [None, []]
-    # This should be:
-    # assert arrow_table["col_list_with_struct"].to_pylist() == [None, [{'test': 1}]]
-    # Once https://github.com/apache/arrow/issues/38809 has been fixed
-    assert arrow_table["col_list_with_struct"].to_pylist() == [[], [{"test": 1}]]
+    assert arrow_table["col_list_with_struct"].to_pylist() == [None, [{"test": 1}]]
 
 
 @pytest.mark.integration
diff --git a/tests/io/test_pyarrow.py b/tests/io/test_pyarrow.py
index b31c18949b..98ccb02ebd 100644
--- a/tests/io/test_pyarrow.py
+++ b/tests/io/test_pyarrow.py
@@ -3210,6 +3210,77 @@ def test__to_requested_schema_float_promotion(
     assert result.column(0).to_pylist() == [1.5, 2.25, 3.0, None]
 
 
+def test__to_requested_schema_null_list_of_structs() -> None:
+    """Test that a null list survives the write path when its element is a struct."""
+    requested_schema = Schema(
+        NestedField(
+            1,
+            "col_list_with_struct",
+            ListType(11, StructType(NestedField(111, "test", IntegerType(), required=False)), element_required=False),
+            required=False,
+        ),
+        NestedField(2, "col_list", ListType(21, IntegerType(), element_required=False), required=False),
+    )
+    file_schema = requested_schema
+
+    arrow_schema = pa.schema(
+        [
+            pa.field("col_list_with_struct", pa.list_(pa.struct([pa.field("test", pa.int32())]))),
+            pa.field("col_list", pa.list_(pa.int32())),
+        ]
+    )
+    batch = pa.RecordBatch.from_arrays(
+        [
+            pa.array([[{"test": 1}], [], None], type=arrow_schema.field(0).type),
+            pa.array([[1], [], None], type=arrow_schema.field(1).type),
+        ],
+        schema=arrow_schema,
+    )
+
+    result = _to_requested_schema(
+        requested_schema, file_schema, batch, downcast_ns_timestamp_to_us=False, include_field_ids=False
+    )
+
+    # A null list and an empty list are different values, and only the struct-element
+    # case ever collapsed the former into the latter.
+    assert result.column(0).to_pylist() == [[{"test": 1}], [], None]
+    assert result.column(1).to_pylist() == [[1], [], None]
+
+
+def test__to_requested_schema_renamed_field_in_list_of_structs() -> None:
+    """Test that a field renamed inside a list element keeps its values."""
+    file_schema = Schema(
+        NestedField(
+            1,
+            "col",
+            ListType(11, StructType(NestedField(111, "before", IntegerType(), required=False)), element_required=False),
+            required=False,
+        ),
+    )
+    requested_schema = Schema(
+        NestedField(
+            1,
+            "col",
+            ListType(11, StructType(NestedField(111, "after", IntegerType(), required=False)), element_required=False),
+            required=False,
+        ),
+    )
+
+    arrow_schema = pa.schema([pa.field("col", pa.list_(pa.struct([pa.field("before", pa.int32())])))])
+    batch = pa.RecordBatch.from_arrays(
+        [pa.array([[{"before": 1}], [], None, [{"before": 3}]], type=arrow_schema.field(0).type)],
+        schema=arrow_schema,
+    )
+
+    result = _to_requested_schema(
+        requested_schema, file_schema, batch, downcast_ns_timestamp_to_us=False, include_field_ids=False
+    )
+
+    # Rebuilding the list from the projected element array is what carries the rename.
+    # Casting instead would match the element fields by name and yield nulls throughout.
+    assert result.column(0).to_pylist() == [[{"after": 1}], [], None, [{"after": 3}]]
+
+
 def test_pyarrow_file_io_fs_by_scheme_cache() -> None:
     # It's better to set up multi-region minio servers for an integration test once `endpoint_url` argument
     # becomes available for `resolve_s3_region`