diff --git a/dask_geopandas/core.py b/dask_geopandas/core.py
index 25e2c6d..b8b9ab7 100644
--- a/dask_geopandas/core.py
+++ b/dask_geopandas/core.py
@@ -389,7 +389,7 @@ def hilbert_distance(self, total_bounds=None, level=16):
             _hilbert_distance,
             total_bounds=total_bounds,
             level=level,
-            meta=pd.Series([], name="hilbert_distance", dtype="uint32"),
+            meta=pd.Series([], name="hilbert_distance", dtype=np.int64),
         )
 
         return distances
diff --git a/dask_geopandas/hilbert_distance.py b/dask_geopandas/hilbert_distance.py
index b8d5a75..8bb597e 100644
--- a/dask_geopandas/hilbert_distance.py
+++ b/dask_geopandas/hilbert_distance.py
@@ -43,7 +43,9 @@ def _hilbert_distance(gdf, total_bounds=None, level=16):
     # Compute distance along hilbert curve
     distances = _encode(level, x, y)
 
-    return pd.Series(distances, index=gdf.index, name="hilbert_distance")
+    return pd.Series(
+        distances, index=gdf.index, name="hilbert_distance", dtype=np.int64
+    )
 
 
 def _continuous_to_discrete_coords(bounds, level, total_bounds):
diff --git a/dask_geopandas/tests/test_spatial_partitioning.py b/dask_geopandas/tests/test_spatial_partitioning.py
index 7c7d8a3..f33ec81 100644
--- a/dask_geopandas/tests/test_spatial_partitioning.py
+++ b/dask_geopandas/tests/test_spatial_partitioning.py
@@ -1,9 +1,12 @@
+import numpy as np
 import pytest
 
 import geopandas
 from geopandas.testing import assert_geodataframe_equal, assert_geoseries_equal
+from shapely.geometry import Point
 
 import dask_geopandas
+from dask_geopandas.hilbert_distance import _hilbert_distance
 
 
 def test_propagate_on_geometry_access():
@@ -60,3 +63,16 @@ def test_cx():
     assert len(subset) == 0
     expected = df.cx[-200:-190, 300:400]
     assert_geodataframe_equal(subset.compute(), expected)
+
+
+def test_geopandas_handles_large_hilbert_distances():
+    df = geopandas.GeoDataFrame(
+        {"geometry": [Point(-103152.516, -8942.156), Point(118914.500, 1010032.562)]}
+    )
+
+    # make sure we have values greater than 32bits
+    dist = _hilbert_distance(df)
+    assert ((dist > np.iinfo(np.int32).max) | (dist < np.iinfo(np.int32).min)).any()
+
+    ddf = dask_geopandas.from_geopandas(df, npartitions=1)
+    ddf.spatial_shuffle()