diff --git a/test/quantization/test_quantized_op.py b/test/quantization/test_quantized_op.py
index 3d87e14a6b72..70bbbbe496b8 100644
--- a/test/quantization/test_quantized_op.py
+++ b/test/quantization/test_quantized_op.py
@@ -4117,7 +4117,7 @@ def test_reflection_pad1d(self, batch_size, channels, width, qtype):
            width=st.integers(16, 128),
            qtype=st.sampled_from(hu._ALL_QINT_TYPES))
     def test_reflection_pad2d(self, batch_size, channels, height, width, qtype):
-        padding = width // 4
+        padding = (width // 4, width // 4, height // 4, height // 4)
 
         x = torch.arange(batch_size * channels * height * width).to(torch.float)
         x = x.resize(batch_size, channels, height, width)