diff --git a/doc/array.rst b/doc/array.rst
index adbd9cfb08bf9a9f6f3db46c15efc5d986d3d237..b10c3efc9bcb5a77c6ec45e1425c32fd97ecc574 100644
--- a/doc/array.rst
+++ b/doc/array.rst
@@ -292,7 +292,7 @@ Constructing :class:`GPUArray` Instances
 
     A synonym for the :class:`GPUArray` constructor.
 
-.. function:: zeros(shape, dtype, *, allocator=None, order="C")
+.. function:: zeros(shape, dtype=np.float64, *, allocator=None, order="C")
 
     Same as :func:`empty`, but the :class:`GPUArray` is zero-initialized before
     being returned.
diff --git a/pycuda/gpuarray.py b/pycuda/gpuarray.py
index 316715d82d6daa89c65515614e246ca917753612..100d21cc7a75325d93295bdc8a89d0bf3e439b01 100644
--- a/pycuda/gpuarray.py
+++ b/pycuda/gpuarray.py
@@ -1306,7 +1306,7 @@ def to_gpu_async(ary, allocator=drv.mem_alloc, stream=None):
 empty = GPUArray
 
 
-def zeros(shape, dtype, allocator=drv.mem_alloc, order="C"):
+def zeros(shape, dtype=np.float64, allocator=drv.mem_alloc, order="C"):
     """Returns an array of the given shape and dtype filled with 0's."""
     result = GPUArray(shape, dtype, allocator, order=order)
     zero = np.zeros((), dtype)
diff --git a/test/test_gpuarray.py b/test/test_gpuarray.py
index 881ac6e99dce782bbb3a99970dc0ac46ac3b96d4..7091fee9481c52f4109ea1085a0d2cb0fa9765dc 100644
--- a/test/test_gpuarray.py
+++ b/test/test_gpuarray.py
@@ -1417,6 +1417,12 @@ class TestGPUArray:
         a[...] = 1729
         np.testing.assert_allclose(a.get(), 1729)
 
+    def test_default_zero(self):
+        # This test was added to make sure that
+        # gpurray.zeros is using np.float64 as the default dtype arg
+        a_gpu = gpuarray.zeros(10)
+        assert a_gpu.dtype == np.float64
+
     @pytest.mark.parametrize("dtype,rtol", [(np.complex64, 1e-6),
                                             (np.complex128, 1e-14)])
     def test_log10(self, dtype, rtol):