diff --git a/databricks/koalas/indexes.py b/databricks/koalas/indexes.py
index 951a5a48dc..1d54e3afe6 100644
--- a/databricks/koalas/indexes.py
+++ b/databricks/koalas/indexes.py
@@ -53,6 +53,7 @@
 from databricks.koalas.series import Series, first_series
 from databricks.koalas.spark.accessors import SparkIndexMethods
 from databricks.koalas.utils import (
+    combine_frames,
     compare_disallow_null,
     default_session,
     is_name_like_tuple,
@@ -3318,6 +3319,40 @@ def item(self) -> Tuple[Scalar, ...]:
         """
         return self._kdf.head(2)._to_internal_pandas().index.item()
 
+    def equal_levels(self, other):
+        """
+        Return True if the levels of both MultiIndex objects are the same
+
+        Examples
+        --------
+        >>> from databricks.koalas.config import set_option, reset_option
+        >>> set_option("compute.ops_on_diff_frames", True)
+
+        >>> kmidx1 = ks.MultiIndex.from_tuples([("a", "x"), ("b", "y"), ("c", "z")])
+        >>> kmidx2 = ks.MultiIndex.from_tuples([("b", "y"), ("a", "x"), ("c", "z")])
+        >>> kmidx1.equal_levels(kmidx2)
+        True
+
+        >>> kmidx2 = ks.MultiIndex.from_tuples([("a", "x"), ("b", "y"), ("c", "j")])
+        >>> kmidx1.equal_levels(kmidx2)
+        False
+
+        >>> reset_option("compute.ops_on_diff_frames")
+        """
+        nlevels = self.nlevels
+        if nlevels != other.nlevels:
+            return False
+        self_frame = self.sort_values().to_frame()
+        other_frame = other.sort_values().to_frame()
+        with option_context("compute.ops_on_diff_frames", True):
+            combined = combine_frames(self_frame, other_frame)
+
+        sdf = combined._internal.spark_frame
+        that_index_name = combined["that"]._internal.data_spark_column_names[0]
+        that_index_scol = scol_for(sdf, that_index_name)
+
+        return len(sdf.filter(that_index_scol.isNull()).head(1)) == 0
+
     def intersection(self, other) -> "MultiIndex":
         """
         Form the intersection of two Index objects.
diff --git a/databricks/koalas/missing/indexes.py b/databricks/koalas/missing/indexes.py
index 0928d3b212..60ea6171e1 100644
--- a/databricks/koalas/missing/indexes.py
+++ b/databricks/koalas/missing/indexes.py
@@ -103,7 +103,6 @@ class MissingPandasLikeMultiIndex(object):
     # Functions
     argsort = _unsupported_function("argsort")
     asof_locs = _unsupported_function("asof_locs")
-    equal_levels = _unsupported_function("equal_levels")
     factorize = _unsupported_function("factorize")
     format = _unsupported_function("format")
     get_indexer = _unsupported_function("get_indexer")
diff --git a/databricks/koalas/tests/test_indexes.py b/databricks/koalas/tests/test_indexes.py
index 32af3f25c4..69c724cbde 100644
--- a/databricks/koalas/tests/test_indexes.py
+++ b/databricks/koalas/tests/test_indexes.py
@@ -1889,6 +1889,29 @@ def test_multiindex_is_unique(self):
 
             self.assertEqual(kdf.index.is_unique, expected)
 
+    def test_multiindex_equal_levels(self):
+        pmidx1 = pd.MultiIndex.from_tuples([("a", "x"), ("b", "y"), ("c", "z")])
+        pmidx2 = pd.MultiIndex.from_tuples([("b", "y"), ("a", "x"), ("c", "z")])
+        kmidx1 = ks.from_pandas(pmidx1)
+        kmidx2 = ks.from_pandas(pmidx2)
+        self.assert_eq(pmidx1.equal_levels(pmidx2), kmidx1.equal_levels(kmidx2))
+
+        pmidx2 = pd.MultiIndex.from_tuples([("a", "x"), ("b", "y"), ("c", "j")])
+        kmidx2 = ks.from_pandas(pmidx2)
+        self.assert_eq(pmidx1.equal_levels(pmidx2), kmidx1.equal_levels(kmidx2))
+
+        pmidx2 = pd.MultiIndex.from_tuples([("a", "x"), ("b", "y"), ("a", "x")])
+        kmidx2 = ks.from_pandas(pmidx2)
+        self.assert_eq(pmidx1.equal_levels(pmidx2), kmidx1.equal_levels(kmidx2))
+
+        pmidx2 = pd.MultiIndex.from_tuples([("a", "x"), ("b", "y")])
+        kmidx2 = ks.from_pandas(pmidx2)
+        self.assert_eq(pmidx1.equal_levels(pmidx2), kmidx1.equal_levels(kmidx2))
+
+        pmidx2 = pd.MultiIndex.from_tuples([("a", "x", "q"), ("b", "y", "w"), ("c", "z", "e")])
+        kmidx2 = ks.from_pandas(pmidx2)
+        self.assert_eq(pmidx1.equal_levels(pmidx2), kmidx1.equal_levels(kmidx2))
+
     def test_view(self):
         pidx = pd.Index([1, 2, 3, 4], name="Koalas")
         kidx = ks.from_pandas(pidx)
diff --git a/docs/source/reference/indexing.rst b/docs/source/reference/indexing.rst
index cb14fe3aec..f7484e6138 100644
--- a/docs/source/reference/indexing.rst
+++ b/docs/source/reference/indexing.rst
@@ -218,6 +218,7 @@ MultiIndex Modifying and computations
    :toctree: api/
 
    MultiIndex.equals
+   MultiIndex.equal_levels
    MultiIndex.identical
    MultiIndex.insert
    MultiIndex.drop