diff --git a/doc/source/release.rst b/doc/source/release.rst
index 4d628fac78cf0..0ef2c29af8139 100644
--- a/doc/source/release.rst
+++ b/doc/source/release.rst
@@ -277,6 +277,7 @@ API Changes
     - ``numexpr`` 2.2.2 fixes incompatiblity in PyTables 2.4 (:issue:`4908`)
     - ``flush`` now accepts an ``fsync`` parameter, which defaults to ``False``
       (:issue:`5364`)
+    - ``unicode`` indices not supported on ``table`` formats (:issue:`5386`)
   - ``JSON``
 
     - added ``date_unit`` parameter to specify resolution of timestamps.
diff --git a/pandas/io/pytables.py b/pandas/io/pytables.py
index 5919589978903..97dc8dcdec73a 100644
--- a/pandas/io/pytables.py
+++ b/pandas/io/pytables.py
@@ -1919,6 +1919,10 @@ def set_version(self):
     def pandas_type(self):
         return _ensure_decoded(getattr(self.group._v_attrs, 'pandas_type', None))
 
+    @property
+    def format_type(self):
+        return 'fixed'
+
     def __unicode__(self):
         """ return a pretty representation of myself """
         self.infer_axes()
@@ -2146,7 +2150,8 @@ def write_index(self, key, index):
             self.write_sparse_intindex(key, index)
         else:
             setattr(self.attrs, '%s_variety' % key, 'regular')
-            converted = _convert_index(index, self.encoding).set_name('index')
+            converted = _convert_index(index, self.encoding,
+                                       self.format_type).set_name('index')
             self.write_array(key, converted.values)
             node = getattr(self.group, key)
             node._v_attrs.kind = converted.kind
@@ -2192,7 +2197,8 @@ def write_multi_index(self, key, index):
                                                  index.names)):
             # write the level
             level_key = '%s_level%d' % (key, i)
-            conv_level = _convert_index(lev, self.encoding).set_name(level_key)
+            conv_level = _convert_index(lev, self.encoding,
+                                        self.format_type).set_name(level_key)
             self.write_array(level_key, conv_level.values)
             node = getattr(self.group, level_key)
             node._v_attrs.kind = conv_level.kind
@@ -2609,6 +2615,10 @@ def __init__(self, *args, **kwargs):
     def table_type_short(self):
         return self.table_type.split('_')[0]
 
+    @property
+    def format_type(self):
+        return 'table'
+
     def __unicode__(self):
         """ return a pretty representatgion of myself """
         self.infer_axes()
@@ -2991,7 +3001,7 @@ def create_axes(self, axes, obj, validate=True, nan_rep=None,
             if i in axes:
                 name = obj._AXIS_NAMES[i]
                 index_axes_map[i] = _convert_index(
-                    a, self.encoding).set_name(name).set_axis(i)
+                    a, self.encoding, self.format_type).set_name(name).set_axis(i)
             else:
 
                 # we might be able to change the axes on the appending data if
@@ -3823,7 +3833,7 @@ def _get_info(info, name):
         idx = info[name] = dict()
     return idx
 
-def _convert_index(index, encoding=None):
+def _convert_index(index, encoding=None, format_type=None):
     index_name = getattr(index, 'name', None)
 
     if isinstance(index, DatetimeIndex):
@@ -3870,9 +3880,13 @@ def _convert_index(index, encoding=None):
             converted, 'string', _tables().StringCol(itemsize), itemsize=itemsize,
                         index_name=index_name)
     elif inferred_type == 'unicode':
-        atom = _tables().ObjectAtom()
-        return IndexCol(np.asarray(values, dtype='O'), 'object', atom,
-                        index_name=index_name)
+        if format_type == 'fixed':
+            atom = _tables().ObjectAtom()
+            return IndexCol(np.asarray(values, dtype='O'), 'object', atom,
+                            index_name=index_name)
+        raise TypeError(
+            "[unicode] is not supported as a in index type for [{0}] formats".format(format_type))
+
     elif inferred_type == 'integer':
         # take a guess for now, hope the values fit
         atom = _tables().Int64Col()
diff --git a/pandas/io/tests/test_pytables.py b/pandas/io/tests/test_pytables.py
index a08073bd7bd35..598f374e0fcf7 100644
--- a/pandas/io/tests/test_pytables.py
+++ b/pandas/io/tests/test_pytables.py
@@ -755,6 +755,38 @@ def test_append_series(self):
             store.append('mi', s)
             tm.assert_series_equal(store['mi'], s)
 
+    def test_store_index_types(self):
+        # GH5386
+        # test storing various index types
+
+        with ensure_clean(self.path) as store:
+
+            def check(format,index):
+                df = DataFrame(np.random.randn(10,2),columns=list('AB'))
+                df.index = index(len(df))
+
+                _maybe_remove(store, 'df')
+                store.put('df',df,format=format)
+                assert_frame_equal(df,store['df'])
+
+            for index in [ tm.makeFloatIndex, tm.makeStringIndex, tm.makeIntIndex,
+                           tm.makeDateIndex, tm.makePeriodIndex ]:
+
+                check('table',index)
+                check('fixed',index)
+
+            # unicode
+            index = tm.makeUnicodeIndex
+            if compat.PY3:
+                check('table',index)
+                check('fixed',index)
+            else:
+
+                # only support for fixed types (and they have a perf warning)
+                self.assertRaises(TypeError, check, 'table', index)
+                with tm.assert_produces_warning(expected_warning=PerformanceWarning):
+                    check('fixed',index)
+
     def test_encoding(self):
 
         if sys.byteorder != 'little':
diff --git a/pandas/util/testing.py b/pandas/util/testing.py
index f40a8e1a5a9d6..2e4d1f3e8df74 100644
--- a/pandas/util/testing.py
+++ b/pandas/util/testing.py
@@ -336,7 +336,8 @@ def ensure_clean(filename=None, return_filelike=False):
             yield filename
         finally:
             try:
-                os.remove(filename)
+                if os.path.exists(filename):
+                    os.remove(filename)
             except Exception as e:
                 print(e)