diff --git a/nnunetv2/experiment_planning/verify_dataset_integrity.py b/nnunetv2/experiment_planning/verify_dataset_integrity.py
index 5a25840f8..1359a9769 100644
--- a/nnunetv2/experiment_planning/verify_dataset_integrity.py
+++ b/nnunetv2/experiment_planning/verify_dataset_integrity.py
@@ -58,6 +58,13 @@ def check_cases(image_files: List[str], label_file: str, expected_num_channels:
         print(f'Segmentation contains NaN pixel values. You need to fix that.\nSegmentation:\n{label_file}')
         ret = False
 
+    # check segmentation is single-channel
+    if segmentation.shape[0] != 1:
+        print('Error: Segmentation must be a single-channel integer label map but has %d channels. '
+              'Please re-save your label file as a single-channel image with integer class indices '
+              '(0, 1, 2, ...). \nSeg file: %s\n' % (segmentation.shape[0], label_file))
+        ret = False
+
     # check shapes
     shape_image = images.shape[1:]
     shape_seg = segmentation.shape[1:]
diff --git a/nnunetv2/imageio/natural_image_reader_writer.py b/nnunetv2/imageio/natural_image_reader_writer.py
index cbf8e614d..57abc5939 100644
--- a/nnunetv2/imageio/natural_image_reader_writer.py
+++ b/nnunetv2/imageio/natural_image_reader_writer.py
@@ -59,7 +59,15 @@ def read_images(self, image_fnames: Union[List[str], Tuple[str, ...]]) -> Tuple[
         return np.vstack(images, dtype=np.float32, casting='unsafe'), {'spacing': (999, 1, 1)}
 
     def read_seg(self, seg_fname: str) -> Tuple[np.ndarray, dict]:
-        return self.read_images((seg_fname, ))
+        seg, props = self.read_images((seg_fname, ))
+        if seg.shape[0] != 1:
+            raise RuntimeError(
+                f"Segmentation file {seg_fname} has {seg.shape[0]} channels. nnU-Net expects "
+                f"segmentation files to be single-channel integer label maps. Please re-save your "
+                f"label files as grayscale (single-channel) images with integer class indices "
+                f"(0, 1, 2, ...)."
+            )
+        return seg, props
 
     def write_seg(self, seg: np.ndarray, output_fname: str, properties: dict) -> None:
         io.imsave(output_fname, seg[0].astype(np.uint8 if np.max(seg) < 255 else np.uint16, copy=False), check_contrast=False)