diff --git a/doc/source/whatsnew/v1.0.0.rst b/doc/source/whatsnew/v1.0.0.rst
index fd1c1271a5e37..dd96c6b594cea 100644
--- a/doc/source/whatsnew/v1.0.0.rst
+++ b/doc/source/whatsnew/v1.0.0.rst
@@ -344,6 +344,7 @@ Reshaping
 - Bug :func:`merge_asof` could not use :class:`datetime.timedelta` for ``tolerance`` kwarg (:issue:`28098`)
 - Bug in :func:`merge`, did not append suffixes correctly with MultiIndex (:issue:`28518`)
 - :func:`qcut` and :func:`cut` now handle boolean input (:issue:`20303`)
+- Fix to ensure all int dtypes can be used in :func:`merge_asof` when using a tolerance value. Previously every non-int64 type would raise an erroneous ``MergeError`` (:issue:`28870`).
 
 Sparse
 ^^^^^^
diff --git a/pandas/core/reshape/merge.py b/pandas/core/reshape/merge.py
index 910c7ea561929..7bfc8153da568 100644
--- a/pandas/core/reshape/merge.py
+++ b/pandas/core/reshape/merge.py
@@ -28,7 +28,6 @@
     is_dtype_equal,
     is_extension_array_dtype,
     is_float_dtype,
-    is_int64_dtype,
     is_integer,
     is_integer_dtype,
     is_list_like,
@@ -1641,7 +1640,7 @@ def _get_merge_keys(self):
                 if self.tolerance < Timedelta(0):
                     raise MergeError("tolerance must be positive")
 
-            elif is_int64_dtype(lt):
+            elif is_integer_dtype(lt):
                 if not is_integer(self.tolerance):
                     raise MergeError(msg)
                 if self.tolerance < 0:
diff --git a/pandas/tests/reshape/merge/test_merge_asof.py b/pandas/tests/reshape/merge/test_merge_asof.py
index caf2539a9e150..2e9ae80323159 100644
--- a/pandas/tests/reshape/merge/test_merge_asof.py
+++ b/pandas/tests/reshape/merge/test_merge_asof.py
@@ -1287,3 +1287,19 @@ def test_timedelta_tolerance_nearest(self):
         )
 
         assert_frame_equal(result, expected)
+
+    def test_int_type_tolerance(self, any_int_dtype):
+        # GH #28870
+
+        left = pd.DataFrame({"a": [0, 10, 20], "left_val": [1, 2, 3]})
+        right = pd.DataFrame({"a": [5, 15, 25], "right_val": [1, 2, 3]})
+        left["a"] = left["a"].astype(any_int_dtype)
+        right["a"] = right["a"].astype(any_int_dtype)
+
+        expected = pd.DataFrame(
+            {"a": [0, 10, 20], "left_val": [1, 2, 3], "right_val": [np.nan, 1.0, 2.0]}
+        )
+        expected["a"] = expected["a"].astype(any_int_dtype)
+
+        result = pd.merge_asof(left, right, on="a", tolerance=10)
+        assert_frame_equal(result, expected)