diff --git a/doc/source/whatsnew/v0.24.0.rst b/doc/source/whatsnew/v0.24.0.rst
index 9209536ff13f9..29ab51c582a97 100644
--- a/doc/source/whatsnew/v0.24.0.rst
+++ b/doc/source/whatsnew/v0.24.0.rst
@@ -1448,6 +1448,7 @@ Conversion
 
 - Bug in :meth:`DataFrame.combine_first` in which column types were unexpectedly converted to float (:issue:`20699`)
 - Bug in :meth:`DataFrame.clip` in which column types are not preserved and casted to float (:issue:`24162`)
+- Bug in :meth:`DataFrame.clip` when order of columns of dataframes doesn't match, result observed is wrong in numeric values (:issue:`20911`)
 
 Strings
 ^^^^^^^
diff --git a/pandas/core/generic.py b/pandas/core/generic.py
index a8d5e4aa772cc..eb14a26e75a9c 100644
--- a/pandas/core/generic.py
+++ b/pandas/core/generic.py
@@ -7176,7 +7176,7 @@ def _clip_with_one_bound(self, threshold, method, axis, inplace):
             if isinstance(self, ABCSeries):
                 threshold = pd.Series(threshold, index=self.index)
             else:
-                threshold = _align_method_FRAME(self, np.asarray(threshold),
+                threshold = _align_method_FRAME(self, threshold,
                                                 axis)
         return self.where(subset, threshold, axis=axis, inplace=inplace)
 
diff --git a/pandas/tests/frame/test_analytics.py b/pandas/tests/frame/test_analytics.py
index baf763d7b1d03..6f68828b94a84 100644
--- a/pandas/tests/frame/test_analytics.py
+++ b/pandas/tests/frame/test_analytics.py
@@ -1971,6 +1971,22 @@ def test_clip_against_frame(self, axis):
         tm.assert_frame_equal(clipped_df[ub_mask], ub[ub_mask])
         tm.assert_frame_equal(clipped_df[mask], df[mask])
 
+    def test_clip_against_unordered_columns(self):
+        # GH 20911
+        df1 = DataFrame(np.random.randn(1000, 4), columns=['A', 'B', 'C', 'D'])
+        df2 = DataFrame(np.random.randn(1000, 4), columns=['D', 'A', 'B', 'C'])
+        df3 = DataFrame(df2.values - 1, columns=['B', 'D', 'C', 'A'])
+        result_upper = df1.clip(lower=0, upper=df2)
+        expected_upper = df1.clip(lower=0, upper=df2[df1.columns])
+        result_lower = df1.clip(lower=df3, upper=3)
+        expected_lower = df1.clip(lower=df3[df1.columns], upper=3)
+        result_lower_upper = df1.clip(lower=df3, upper=df2)
+        expected_lower_upper = df1.clip(lower=df3[df1.columns],
+                                        upper=df2[df1.columns])
+        tm.assert_frame_equal(result_upper, expected_upper)
+        tm.assert_frame_equal(result_lower, expected_lower)
+        tm.assert_frame_equal(result_lower_upper, expected_lower_upper)
+
     def test_clip_with_na_args(self, float_frame):
         """Should process np.nan argument as None """
         # GH 17276