diff --git a/pandas/core/common.py b/pandas/core/common.py
index 1de8269c9a0c6..ec516d9d80023 100644
--- a/pandas/core/common.py
+++ b/pandas/core/common.py
@@ -410,19 +410,6 @@ def _apply_if_callable(maybe_callable, obj, **kwargs):
     return maybe_callable
 
 
-def _where_compat(mask, arr1, arr2):
-    if arr1.dtype == _NS_DTYPE and arr2.dtype == _NS_DTYPE:
-        new_vals = np.where(mask, arr1.view('i8'), arr2.view('i8'))
-        return new_vals.view(_NS_DTYPE)
-
-    if arr1.dtype == _NS_DTYPE:
-        arr1 = tslib.ints_to_pydatetime(arr1.view('i8'))
-    if arr2.dtype == _NS_DTYPE:
-        arr2 = tslib.ints_to_pydatetime(arr2.view('i8'))
-
-    return np.where(mask, arr1, arr2)
-
-
 def _dict_compat(d):
     """
     Helper function to convert datetimelike-keyed dicts to Timestamp-keyed dict
diff --git a/pandas/core/series.py b/pandas/core/series.py
index 0450f28087f66..801a8ddeb67f7 100644
--- a/pandas/core/series.py
+++ b/pandas/core/series.py
@@ -29,13 +29,16 @@
     is_hashable,
     is_iterator,
     is_dict_like,
+    is_dtype_equal,
     is_scalar,
     _is_unorderable_exception,
     _ensure_platform_int,
-    pandas_dtype)
+    pandas_dtype,
+    needs_i8_conversion)
 from pandas.core.dtypes.generic import (
     ABCSparseArray, ABCDataFrame, ABCIndexClass)
 from pandas.core.dtypes.cast import (
+    find_common_type, maybe_downcast_to_dtype,
     maybe_upcast, infer_dtype_from_scalar,
     maybe_convert_platform,
     maybe_cast_to_datetime, maybe_castable,
@@ -2304,7 +2307,24 @@ def combine_first(self, other):
         other = other.reindex(new_index, copy=False)
         # TODO: do we need name?
         name = ops.get_op_result_name(self, other)  # noqa
-        rs_vals = com._where_compat(isna(this), other._values, this._values)
+        if not is_dtype_equal(this.dtype, other.dtype):
+            new_dtype = find_common_type([this.dtype, other.dtype])
+            if not is_dtype_equal(this.dtype, new_dtype):
+                this = this.astype(new_dtype)
+            if not is_dtype_equal(other.dtype, new_dtype):
+                other = other.astype(new_dtype)
+
+        if needs_i8_conversion(this.dtype):
+            mask = isna(this)
+            this_values = this.values.view('i8')
+            other_values = other.values.view('i8')
+        else:
+            this_values = this.values
+            other_values = other.values
+            mask = isna(this_values)
+
+        rs_vals = np.where(mask, other_values, this_values)
+        rs_vals = maybe_downcast_to_dtype(rs_vals, this.dtype)
         return self._constructor(rs_vals, index=new_index).__finalize__(self)
 
     def update(self, other):
diff --git a/pandas/tests/series/test_combine_concat.py b/pandas/tests/series/test_combine_concat.py
index f35cce6ac9d71..d3e4720a756f1 100644
--- a/pandas/tests/series/test_combine_concat.py
+++ b/pandas/tests/series/test_combine_concat.py
@@ -170,6 +170,19 @@ def get_result_type(dtype, dtype2):
                                     ]).dtype
                 assert result.kind == expected
 
+    def test_combine_first_dt_tz_values(self):
+        dts1 = pd.date_range('20150101', '20150105', tz='America/New_York')
+        df1 = pd.DataFrame({'date': dts1})
+        dts2 = pd.date_range('20160514', '20160518', tz='America/New_York')
+        df2 = pd.DataFrame({'date': dts2}, index=range(3, 8))
+        result = df1.date.combine_first(df2.date)
+        exp_vals = pd.DatetimeIndex(['20150101', '20150102', '20150103',
+                                     '20150104', '20150105', '20160516',
+                                     '20160517', '20160518'],
+                                    tz='America/New_York')
+        exp = pd.Series(exp_vals, name='date')
+        assert_series_equal(exp, result)
+
     def test_concat_empty_series_dtypes(self):
 
         # booleans