diff --git a/pandas/core/index.py b/pandas/core/index.py
index 2252ba666ca59..6cbffd5b4dbba 100644
--- a/pandas/core/index.py
+++ b/pandas/core/index.py
@@ -148,6 +148,9 @@ def __new__(cls, data, dtype=None, copy=False, name=None, fastpath=False,
             if copy:
                 subarr = subarr.copy()
 
+        elif hasattr(data, '__array__'):
+            return Index(np.asarray(data), dtype=dtype, copy=copy, name=name,
+                         **kwargs)
         elif np.isscalar(data):
             cls._scalar_data_error(data)
         else:
diff --git a/pandas/tests/test_index.py b/pandas/tests/test_index.py
index f2372f98b330b..203d3f26d0812 100644
--- a/pandas/tests/test_index.py
+++ b/pandas/tests/test_index.py
@@ -177,12 +177,17 @@ def test_constructor_ndarray_like(self):
         # it should be possible to convert any object that satisfies the numpy
         # ndarray interface directly into an Index
         class ArrayLike(object):
+            def __init__(self, array):
+                self.array = array
             def __array__(self, dtype=None):
-                return np.arange(5)
+                return self.array
 
-        expected = pd.Index(np.arange(5))
-        result = pd.Index(ArrayLike())
-        self.assertTrue(result.equals(expected))
+        for array in [np.arange(5),
+                      np.array(['a', 'b', 'c']),
+                      pd.date_range('2000-01-01', periods=3).values]:
+            expected = pd.Index(array)
+            result = pd.Index(ArrayLike(array))
+            self.assertTrue(result.equals(expected))
 
     def test_index_ctor_infer_periodindex(self):
         from pandas import period_range, PeriodIndex
@@ -447,7 +452,7 @@ def test_intersection(self):
         assertRaisesRegexp(TypeError, "iterable", first.intersection, 0.5)
 
         idx1 = Index([1, 2, 3, 4, 5], name='idx')
-        # if target has the same name, it is preserved 
+        # if target has the same name, it is preserved
         idx2 = Index([3, 4, 5, 6, 7], name='idx')
         expected2 = Index([3, 4, 5], name='idx')
         result2 = idx1.intersection(idx2)