diff --git a/pandas/core/indexing.py b/pandas/core/indexing.py
index c30885291ffc9..2566f05daeaea 100755
--- a/pandas/core/indexing.py
+++ b/pandas/core/indexing.py
@@ -123,6 +123,9 @@ def __getitem__(self, key):
                 #  generally slice or list.
                 # TODO(ix): most/all of the TypeError cases here are for ix,
                 #  so this check can be removed once ix is removed.
+                # The InvalidIndexError is only catched for compatibility
+                #  with geopandas, see
+                #  https://github.com/pandas-dev/pandas/issues/27258
                 pass
             else:
                 if is_scalar(values):
diff --git a/pandas/tests/test_downstream.py b/pandas/tests/test_downstream.py
index d644c002fbdfb..93baafddedeb4 100644
--- a/pandas/tests/test_downstream.py
+++ b/pandas/tests/test_downstream.py
@@ -10,7 +10,7 @@
 
 from pandas.compat import PY36
 
-from pandas import DataFrame
+from pandas import DataFrame, Series
 from pandas.util import testing as tm
 
 
@@ -123,6 +123,26 @@ def test_geopandas():
     assert geopandas.read_file(fp) is not None
 
 
+def test_geopandas_coordinate_indexer():
+    # this test is included to have coverage of one case in the indexing.py
+    # code that is only kept for compatibility with geopandas, see
+    # https://github.com/pandas-dev/pandas/issues/27258
+    # We should be able to remove this after some time when its usage is
+    # removed in geopandas
+    from pandas.core.indexing import _NDFrameIndexer
+
+    class _CoordinateIndexer(_NDFrameIndexer):
+        def _getitem_tuple(self, tup):
+            obj = self.obj
+            xs, ys = tup
+            return obj[xs][ys]
+
+    Series._create_indexer("cx", _CoordinateIndexer)
+    s = Series(range(5))
+    res = s.cx[:, :]
+    tm.assert_series_equal(s, res)
+
+
 # Cython import warning
 @pytest.mark.filterwarnings("ignore:can't resolve:ImportWarning")
 def test_pyarrow(df):