diff --git a/doc/source/whatsnew/v0.21.1.txt b/doc/source/whatsnew/v0.21.1.txt
index fcc9144bef9e3..0ad0cac0a4181 100644
--- a/doc/source/whatsnew/v0.21.1.txt
+++ b/doc/source/whatsnew/v0.21.1.txt
@@ -120,6 +120,7 @@ I/O
 - Bug in :meth:`DataFrame.to_msgpack` when serializing data of the numpy.bool_ datatype (:issue:`18390`)
 - Bug in :func:`read_json` not decoding when reading line deliminted JSON from S3 (:issue:`17200`)
 - Bug in :func:`pandas.io.json.json_normalize` to avoid modification of ``meta`` (:issue:`18610`)
+- Bug in :func:`to_latex` where repeated multi-index values were not printed even though a higher level index differed from the previous row (:issue:`14484`)
 
 Plotting
 ^^^^^^^^
diff --git a/pandas/io/formats/format.py b/pandas/io/formats/format.py
index 8f25eb3af70cd..97abf60a40a5b 100644
--- a/pandas/io/formats/format.py
+++ b/pandas/io/formats/format.py
@@ -46,7 +46,6 @@
 import pandas as pd
 import numpy as np
 
-import itertools
 import csv
 from functools import partial
 
@@ -903,6 +902,7 @@ def get_col_type(dtype):
             name = any(self.frame.index.names)
             cname = any(self.frame.columns.names)
             lastcol = self.frame.index.nlevels - 1
+            previous_lev3 = None
             for i, lev in enumerate(self.frame.index.levels):
                 lev2 = lev.format()
                 blank = ' ' * len(lev2[0])
@@ -913,11 +913,19 @@ def get_col_type(dtype):
                     lev3 = [blank] * clevels
                 if name:
                     lev3.append(lev.name)
-                for level_idx, group in itertools.groupby(
-                        self.frame.index.labels[i]):
-                    count = len(list(group))
-                    lev3.extend([lev2[level_idx]] + [blank] * (count - 1))
+                current_idx_val = None
+                for level_idx in self.frame.index.labels[i]:
+                    if ((previous_lev3 is None or
+                        previous_lev3[len(lev3)].isspace()) and
+                            lev2[level_idx] == current_idx_val):
+                        # same index as above row and left index was the same
+                        lev3.append(blank)
+                    else:
+                        # different value than above or left index different
+                        lev3.append(lev2[level_idx])
+                        current_idx_val = lev2[level_idx]
                 strcols.insert(i, lev3)
+                previous_lev3 = lev3
 
         column_format = self.column_format
         if column_format is None:
diff --git a/pandas/tests/io/formats/test_to_latex.py b/pandas/tests/io/formats/test_to_latex.py
index 35ef5a1cf5c72..7d42ff20ea31e 100644
--- a/pandas/tests/io/formats/test_to_latex.py
+++ b/pandas/tests/io/formats/test_to_latex.py
@@ -221,6 +221,28 @@ def test_to_latex_multiindex(self):
 
         assert result == expected
 
+    def test_to_latex_multiindex_dupe_level(self):
+        # see gh-14484
+        #
+        # If an index is repeated in subsequent rows, it should be
+        # replaced with a blank in the created table. This should
+        # ONLY happen if all higher order indices (to the left) are
+        # equal too. In this test, 'c' has to be printed both times
+        # because the higher order index 'A' != 'B'.
+        df = pd.DataFrame(index=pd.MultiIndex.from_tuples(
+            [('A', 'c'), ('B', 'c')]), columns=['col'])
+        result = df.to_latex()
+        expected = r"""\begin{tabular}{lll}
+\toprule
+  &   &  col \\
+\midrule
+A & c &  NaN \\
+B & c &  NaN \\
+\bottomrule
+\end{tabular}
+"""
+        assert result == expected
+
     def test_to_latex_multicolumnrow(self):
         df = pd.DataFrame({
             ('c1', 0): {x: x for x in range(5)},