diff --git a/lib/matplotlib/image.py b/lib/matplotlib/image.py
index 6ca472518b65..390f5e95b4ac 100644
--- a/lib/matplotlib/image.py
+++ b/lib/matplotlib/image.py
@@ -423,7 +423,7 @@ def _make_image(self, A, in_bbox, out_bbox, clip_bbox, magnification=1.0,
         # So that the image is aligned with the edge of the Axes, we want to
         # round up the output width to the next integer.  This also means
         # scaling the transform slightly to account for the extra subpixel.
-        if ((not unsampled) and t.is_affine and round_to_pixel_border and
+        if ((not unsampled) and round_to_pixel_border and
                 (out_width_base % 1.0 != 0.0 or out_height_base % 1.0 != 0.0)):
             out_width = math.ceil(out_width_base)
             out_height = math.ceil(out_height_base)
diff --git a/lib/matplotlib/tests/test_image.py b/lib/matplotlib/tests/test_image.py
index 1c89bc5e7912..aa8f6ed835dd 100644
--- a/lib/matplotlib/tests/test_image.py
+++ b/lib/matplotlib/tests/test_image.py
@@ -1641,6 +1641,17 @@ def test__resample_valid_output():
         resample(np.zeros((9, 9)), out)
 
 
+@pytest.fixture
+def nonaffine_identity():
+    class NonAffineIdentityTransform(Transform):
+        input_dims = 2
+        output_dims = 2
+
+        def inverted(self):
+            return self
+    return NonAffineIdentityTransform()
+
+
 @pytest.mark.parametrize("data, interpolation, expected",
     [(np.array([[0.1, 0.3, 0.2]]), mimage.NEAREST,
       np.array([[0.1, 0.1, 0.1, 0.3, 0.3, 0.3, 0.3, 0.2, 0.2, 0.2]])),
@@ -1649,7 +1660,7 @@ def test__resample_valid_output():
                  0.28476562, 0.2546875, 0.22460938, 0.20002441, 0.20002441]])),
     ]
 )
-def test_resample_nonaffine(data, interpolation, expected):
+def test_resample_nonaffine(data, interpolation, expected, nonaffine_identity):
     # Test that equivalent affine and nonaffine transforms resample the same
 
     # Create a simple affine transform for scaling the input array
@@ -1661,13 +1672,7 @@ def test_resample_nonaffine(data, interpolation, expected):
 
     # Create a nonaffine version of the same transform
     # by compositing with a nonaffine identity transform
-    class NonAffineIdentityTransform(Transform):
-        input_dims = 2
-        output_dims = 2
-
-        def inverted(self):
-           return self
-    nonaffine_transform = NonAffineIdentityTransform() + affine_transform
+    nonaffine_transform = nonaffine_identity + affine_transform
 
     nonaffine_result = np.empty_like(expected)
     mimage.resample(data, nonaffine_result, nonaffine_transform,
@@ -1675,6 +1680,28 @@ def inverted(self):
     assert_allclose(nonaffine_result, expected, atol=5e-3)
 
 
+@check_figures_equal()
+def test_nonaffine_scaling_to_axes_edges(fig_test, fig_ref, nonaffine_identity):
+    # Test that plotting an image with equivalent affine and nonaffine
+    # transforms is scaled the same to the axes edges
+    data = np.arange(16).reshape((4, 4)) % 3
+
+    # Specifically choose an axes bbox that has a fractional pixel
+
+    fig_test.set_size_inches(5, 5)
+    fig_test.set_dpi(100)
+    ax = fig_test.subplots()
+    ax.set_position([0.2, 0.2, 300.5 / 500, 300.5 / 500])
+    ax.imshow(data, interpolation='nearest',
+              transform=nonaffine_identity + ax.transData)
+
+    fig_ref.set_size_inches(5, 5)
+    fig_ref.set_dpi(100)
+    ax = fig_ref.subplots()
+    ax.set_position([0.2, 0.2, 300.5 / 500, 300.5 / 500])
+    ax.imshow(data, interpolation='nearest')
+
+
 def test_axesimage_get_shape():
     # generate dummy image to test get_shape method
     ax = plt.gca()