diff --git a/nd2reader/stitched.py b/nd2reader/stitched.py
index 9721e212d31fe30b1c76675e12c11f0aee637476..a2f685065df6b83fa9770e004c7c4e156765998c 100644
--- a/nd2reader/stitched.py
+++ b/nd2reader/stitched.py
@@ -9,8 +9,9 @@ def get_unwanted_bytes_ids(image_group_data, image_data_start, height, width):
     # rows (height), as the same unmber of unwanted bytes is expected to be
     # appended at the end of each row. Then, returns the indexes of the unwanted
     # bytes.
+    # Skip the first 4 elements that correspond to the time stamp
     number_of_true_channels = int(len(image_group_data[4:]) / (height * width))
-    n_unwanted_bytes = (len(image_group_data[image_data_start:])) % (height * width)
+    n_unwanted_bytes = (len(image_group_data[4:])) % (height * width)
     if not n_unwanted_bytes:
         return np.arange(0)
     assert 0 == n_unwanted_bytes % height, (