diff --git a/test/test_datasets.py b/test/test_datasets.py
index 22c14cbc08d..88fb3af343a 100644
--- a/test/test_datasets.py
+++ b/test/test_datasets.py
@@ -3297,6 +3297,16 @@ def test_splits(self):
                 for left, right, disparity in dataset:
                     datasets_utils.shape_test_for_stereo(left, right, disparity)
 
+    def test_disparity_scale(self):
+        with self.create_dataset(split="train") as (dataset, _):
+            disparity_path = dataset._disparities[0][0]
+            disparity_image = PIL.Image.fromarray(
+                np.full((100, 200), 100, dtype=np.uint16)
+            )
+            disparity_image.save(disparity_path)
+            disparity = dataset._read_disparity(disparity_path)[0]
+            np.testing.assert_array_equal(disparity, np.ones((1, 100, 200)))
+
     def test_bad_input(self):
         with pytest.raises(
             ValueError, match="Unknown value 'bad' for argument split. Valid values are {'train', 'test'}."
diff --git a/torchvision/datasets/_stereo_matching.py b/torchvision/datasets/_stereo_matching.py
index bc2236e97b8..5bca961da20 100644
--- a/torchvision/datasets/_stereo_matching.py
+++ b/torchvision/datasets/_stereo_matching.py
@@ -1104,7 +1104,7 @@ def __init__(self, root: Union[str, Path], split: str = "train", transforms: Opt
     def _read_disparity(self, file_path: str) -> tuple[np.ndarray, None]:
         disparity_map = np.asarray(Image.open(file_path), dtype=np.float32)
         # unsqueeze disparity to (C, H, W)
-        disparity_map = disparity_map[None, :, :] / 1024.0
+        disparity_map = disparity_map[None, :, :] / 100.0
         valid_mask = None
         return disparity_map, valid_mask