diff --git a/sklearn/linear_model/huber.py b/sklearn/linear_model/huber.py
index 65c6864007eb2..4792f5d848c0b 100644
--- a/sklearn/linear_model/huber.py
+++ b/sklearn/linear_model/huber.py
@@ -11,6 +11,7 @@
 from ..utils import check_consistent_length
 from ..utils import axis0_safe_slice
 from ..utils.extmath import safe_sparse_dot
+from ..utils.validation import FLOAT_DTYPES
 
 
 def _huber_loss_and_gradient(w, X, y, epsilon, alpha, sample_weight=None):
@@ -251,7 +252,8 @@ def fit(self, X, y, sample_weight=None):
         self : object
         """
         X, y = check_X_y(
-            X, y, copy=False, accept_sparse=['csr'], y_numeric=True)
+            X, y, copy=False, accept_sparse=['csr'],
+            y_numeric=True, dtype=FLOAT_DTYPES)
         if sample_weight is not None:
             sample_weight = np.array(sample_weight)
             check_consistent_length(y, sample_weight)
diff --git a/sklearn/linear_model/tests/test_huber.py b/sklearn/linear_model/tests/test_huber.py
index 6a8b26133d5ac..54f7443cfb56f 100644
--- a/sklearn/linear_model/tests/test_huber.py
+++ b/sklearn/linear_model/tests/test_huber.py
@@ -199,3 +199,40 @@ def test_huber_better_r2_score():
 
     # The huber model should also fit poorly on the outliers.
     assert_greater(ridge_outlier_score, huber_outlier_score)
+
+
+def test_huber_bool_X_dense_cast_and_fit():
+    X, y = make_regression(n_samples=40, n_features=5, noise=2.0,
+                           random_state=0)
+    X_bool = X > 0
+    X_float = X_bool.astype(np.float64)
+
+    huber_bool = HuberRegressor()
+    huber_bool.fit(X_bool, y)
+
+    huber_float = HuberRegressor()
+    huber_float.fit(X_float, y)
+
+    assert_array_almost_equal(huber_bool.coef_, huber_float.coef_)
+    assert_almost_equal(huber_bool.intercept_, huber_float.intercept_)
+    assert_array_almost_equal(huber_bool.predict(X_bool),
+                              huber_float.predict(X_float))
+
+
+def test_huber_bool_X_sparse_cast_and_fit():
+    X, y = make_regression(n_samples=40, n_features=5, noise=2.0,
+                           random_state=0)
+    X_bool = X > 0
+    X_sparse_bool = sparse.csr_matrix(X_bool)
+    X_sparse_float = X_sparse_bool.astype(np.float64)
+
+    huber_bool = HuberRegressor()
+    huber_bool.fit(X_sparse_bool, y)
+
+    huber_float = HuberRegressor()
+    huber_float.fit(X_sparse_float, y)
+
+    assert_array_almost_equal(huber_bool.coef_, huber_float.coef_)
+    assert_almost_equal(huber_bool.intercept_, huber_float.intercept_)
+    assert_array_almost_equal(huber_bool.predict(X_sparse_bool),
+                              huber_float.predict(X_sparse_float))