diff --git a/doc/source/whatsnew/v0.17.0.txt b/doc/source/whatsnew/v0.17.0.txt
index 3e81a923a114c..70a33a6915516 100644
--- a/doc/source/whatsnew/v0.17.0.txt
+++ b/doc/source/whatsnew/v0.17.0.txt
@@ -811,6 +811,7 @@ Bug Fixes
 - Bug in ``read_csv`` when using the ``nrows`` or ``chunksize`` parameters if file contains only a header line (:issue:`9535`)
 - Bug in serialization of ``category`` types in HDF5 in presence of alternate encodings. (:issue:`10366`)
 - Bug in ``pd.DataFrame`` when constructing an empty DataFrame with a string dtype (:issue:`9428`)
+- Bug in ``pd.DataFrame.diff`` when DataFrame is not consolidated (:issue:`10907`)
 - Bug in ``pd.unique`` for arrays with the ``datetime64`` or ``timedelta64`` dtype that meant an array with object dtype was returned instead the original dtype (:issue:`9431`)
 - Bug in ``DatetimeIndex.take`` and ``TimedeltaIndex.take`` may not raise ``IndexError`` against invalid index (:issue:`10295`)
 - Bug in ``Series([np.nan]).astype('M8[ms]')``, which now returns ``Series([pd.NaT])`` (:issue:`10747`)
diff --git a/pandas/core/internals.py b/pandas/core/internals.py
index 1d6269ae904d2..15069bf23672b 100644
--- a/pandas/core/internals.py
+++ b/pandas/core/internals.py
@@ -2414,7 +2414,7 @@ def _verify_integrity(self):
                                  'tot_items: {1}'.format(len(self.items),
                                                          tot_items))
 
-    def apply(self, f, axes=None, filter=None, do_integrity_check=False, **kwargs):
+    def apply(self, f, axes=None, filter=None, do_integrity_check=False, consolidate=True, **kwargs):
         """
         iterate over the blocks, collect and create a new block manager
 
@@ -2425,6 +2425,7 @@ def apply(self, f, axes=None, filter=None, do_integrity_check=False, **kwargs):
         filter : list, if supplied, only call the block if the filter is in
                  the block
         do_integrity_check : boolean, default False. Do the block manager integrity check
+        consolidate: boolean, default True. Join together blocks having same dtype
 
         Returns
         -------
@@ -2443,6 +2444,9 @@ def apply(self, f, axes=None, filter=None, do_integrity_check=False, **kwargs):
             else:
                 kwargs['filter'] = filter_locs
 
+        if consolidate:
+            self._consolidate_inplace()
+
         if f == 'where':
             align_copy = True
             if kwargs.get('align', True):
diff --git a/pandas/tests/test_frame.py b/pandas/tests/test_frame.py
index 58c6d15f8ada5..57a43592b3866 100644
--- a/pandas/tests/test_frame.py
+++ b/pandas/tests/test_frame.py
@@ -10771,6 +10771,14 @@ def test_diff(self):
         assert_series_equal(the_diff['A'],
                             tf['A'] - tf['A'].shift(1))
 
+        # issue 10907
+        df = pd.DataFrame({'y': pd.Series([2]), 'z': pd.Series([3])})
+        df.insert(0, 'x', 1)
+        result = df.diff(axis=1)
+        expected = pd.DataFrame({'x':np.nan, 'y':pd.Series(1), 'z':pd.Series(1)}).astype('float64')
+        self.assert_frame_equal(result, expected)
+
+
     def test_diff_timedelta(self):
         # GH 4533
         df = DataFrame(dict(time=[Timestamp('20130101 9:01'),