diff --git a/doc/source/whatsnew/v1.2.0.rst b/doc/source/whatsnew/v1.2.0.rst
index 84ac2d0c17676..ac7a2ddd477c9 100644
--- a/doc/source/whatsnew/v1.2.0.rst
+++ b/doc/source/whatsnew/v1.2.0.rst
@@ -659,6 +659,7 @@ Indexing
 - Bug in indexing on a :class:`Series` or :class:`DataFrame` with a :class:`MultiIndex` and a level named ``"0"`` (:issue:`37194`)
 - Bug in :meth:`Series.__getitem__` when using an unsigned integer array as an indexer giving incorrect results or segfaulting instead of raising ``KeyError`` (:issue:`37218`)
 - Bug in :meth:`Index.where` incorrectly casting numeric values to strings (:issue:`37591`)
+- Bug in :meth:`DataFrame.loc` returning empty result when indexer is a slice with negative step size (:issue:`38071`)
 - Bug in :meth:`Series.loc` and :meth:`DataFrame.loc` raises when the index was of ``object`` dtype and the given numeric label was in the index (:issue:`26491`)
 - Bug in :meth:`DataFrame.loc` returned requested key plus missing values when ``loc`` was applied to single level from a :class:`MultiIndex` (:issue:`27104`)
 - Bug in indexing on a :class:`Series` or :class:`DataFrame` with a :class:`CategoricalIndex` using a listlike indexer containing NA values (:issue:`37722`)
diff --git a/pandas/core/indexes/multi.py b/pandas/core/indexes/multi.py
index d575c67cb36aa..6af6555007c2f 100644
--- a/pandas/core/indexes/multi.py
+++ b/pandas/core/indexes/multi.py
@@ -3078,8 +3078,11 @@ def convert_indexer(start, stop, step, indexer=indexer, codes=level_codes):
             # given the inputs and the codes/indexer, compute an indexer set
             # if we have a provided indexer, then this need not consider
             # the entire labels set
-
+            if step is not None and step < 0:
+                # Switch elements for negative step size
+                start, stop = stop - 1, start - 1
             r = np.arange(start, stop, step)
+
             if indexer is not None and len(indexer) != len(codes):
 
                 # we have an indexer which maps the locations in the labels
@@ -3342,6 +3345,8 @@ def _reorder_indexer(
                         k_codes = k_codes[k_codes >= 0]  # Filter absent keys
                         # True if the given codes are not ordered
                         need_sort = (k_codes[:-1] > k_codes[1:]).any()
+                elif isinstance(k, slice) and k.step is not None and k.step < 0:
+                    need_sort = True
             # Bail out if both index and seq are sorted
             if not need_sort:
                 return indexer
@@ -3368,6 +3373,8 @@ def _reorder_indexer(
                 key_order_map[level_indexer] = np.arange(len(level_indexer))
 
                 new_order = key_order_map[self.codes[i][indexer]]
+            elif isinstance(k, slice) and k.step is not None and k.step < 0:
+                new_order = np.arange(n)[k][indexer]
             elif isinstance(k, slice) and k.start is None and k.stop is None:
                 # slice(None) should not determine order GH#31330
                 new_order = np.ones((n,))[indexer]
diff --git a/pandas/tests/indexing/multiindex/test_slice.py b/pandas/tests/indexing/multiindex/test_slice.py
index d58bc4713f99f..51684f092aefd 100644
--- a/pandas/tests/indexing/multiindex/test_slice.py
+++ b/pandas/tests/indexing/multiindex/test_slice.py
@@ -779,3 +779,13 @@ def test_non_reducing_slice_on_multiindex(self):
         result = df.loc[tslice_]
         expected = DataFrame({("b", "d"): [4, 1]})
         tm.assert_frame_equal(result, expected)
+
+    def test_loc_slice_negative_stepsize(self):
+        # GH#38071
+        mi = MultiIndex.from_product([["a", "b"], [0, 1]])
+        df = DataFrame([[1, 2], [3, 4], [5, 6], [7, 8]], index=mi)
+        result = df.loc[("a", slice(None, None, -1)), :]
+        expected = DataFrame(
+            [[3, 4], [1, 2]], index=MultiIndex.from_tuples([("a", 1), ("a", 0)])
+        )
+        tm.assert_frame_equal(result, expected)