diff --git a/src/spatialdata_io/readers/xenium.py b/src/spatialdata_io/readers/xenium.py
index 5450d451..fcf2b6e1 100644
--- a/src/spatialdata_io/readers/xenium.py
+++ b/src/spatialdata_io/readers/xenium.py
@@ -605,6 +605,7 @@ def _get_polygons(
         else:
             # Use cell_id (str) as GeoDataFrame index.
             geo_df = GeoDataFrame({"geometry": geoms}, index=indices_mapping["cell_id"].values)
+            geo_df.index.name = "cell_id"
     else:
         # Fall back to extracting unique cell IDs from parquet (slow for large_string columns).
         # Triggered when indices_mapping is None: v < 1.3.0 (both nuclei and cells, because
@@ -617,6 +618,7 @@ def _get_polygons(
         unique_ids = id_col.filter(np.concatenate([[True], change_mask])).to_pylist()
         index = _decode_cell_id_column(pd.Series(unique_ids))
         geo_df = GeoDataFrame({"geometry": geoms}, index=index.values)
+        geo_df.index.name = "cell_id"
 
     scale = Scale([1.0 / specs["pixel_size"], 1.0 / specs["pixel_size"]], axes=("x", "y"))
     return ShapesModel.parse(geo_df, transformations={"global": scale})
diff --git a/tests/test_xenium.py b/tests/test_xenium.py
index be353130..17903888 100644
--- a/tests/test_xenium.py
+++ b/tests/test_xenium.py
@@ -114,6 +114,7 @@ def test_example_data_index_integrity(dataset: str) -> None:
         assert sdata["nucleus_labels"]["scale0"]["image"].sel(y=3515.5, x=4618.5).data.compute() == 6392
         assert np.allclose(sdata['transcripts'].compute().loc[[0, 10000, 1113949]]['x'], [2.608911, 194.917831, 1227.499268])
         assert np.isclose(sdata['cell_boundaries'].loc['oipggjko-1'].geometry.centroid.x,736.4864931162789)
+        assert sdata['cell_boundaries'].index.name == 'cell_id'
         index = sdata['nucleus_boundaries']['cell_id'].index[sdata['nucleus_boundaries']['cell_id'].eq('oipggjko-1')][0]
         assert np.isclose(sdata['nucleus_boundaries'].loc[index].geometry.centroid.x,736.4931256878282)
         assert np.array_equal(sdata['table'].X.indices[:3], [1, 3, 34])
@@ -140,6 +141,7 @@ def test_example_data_index_integrity(dataset: str) -> None:
         assert sdata["nucleus_labels"]["scale0"]["image"].sel(y=18.5, x=3015.5).data.compute() == 2764
         assert np.allclose(sdata['transcripts'].compute().loc[[0, 10000, 20000]]['x'], [174.258392, 12.210024, 214.759186])
         assert np.isclose(sdata['cell_boundaries'].loc['aaanbaof-1'].geometry.centroid.x, 43.96894317275074)
+        assert sdata['cell_boundaries'].index.name == 'cell_id'
         index = sdata['nucleus_boundaries']['cell_id'].index[sdata['nucleus_boundaries']['cell_id'].eq('aaanbaof-1')][0]
         assert np.isclose(sdata['nucleus_boundaries'].loc[index].geometry.centroid.x,43.31874577809517)
         assert np.array_equal(sdata['table'].X.indices[:3], [1, 8, 19])
@@ -167,6 +169,7 @@ def test_example_data_index_integrity(dataset: str) -> None:
         assert sdata["nucleus_labels"]["scale0"]["image"].sel(y=4039.5, x=93.5).data.compute() == 274
         assert np.allclose(sdata['transcripts'].compute().loc[[0, 10000, 20000]]['x'], [43.296875, 62.484375, 93.125])
         assert np.isclose(sdata['cell_boundaries'].loc['aadmbfof-1'].geometry.centroid.x, 64.54541104696033)
+        assert sdata['cell_boundaries'].index.name == 'cell_id'
         index = sdata['nucleus_boundaries']['cell_id'].index[sdata['nucleus_boundaries']['cell_id'].eq('aadmbfof-1')][0]
         assert np.isclose(sdata['nucleus_boundaries'].loc[index].geometry.centroid.x, 65.43305896114295)
         assert np.array_equal(sdata['table'].X.indices[:3], [3, 49, 53])