diff --git a/paddle/phi/kernels/cpu/sign_kernel.cc b/paddle/phi/kernels/cpu/sign_kernel.cc
index f03f39f80dcbe..0db47fc2f5266 100644
--- a/paddle/phi/kernels/cpu/sign_kernel.cc
+++ b/paddle/phi/kernels/cpu/sign_kernel.cc
@@ -25,6 +25,7 @@ PD_REGISTER_KERNEL(sign,
                    CPU,
                    ALL_LAYOUT,
                    phi::SignKernel,
+                   uint8_t,
                    int8_t,
                    int16_t,
                    int32_t,
diff --git a/paddle/phi/kernels/funcs/eigen/sign.cc b/paddle/phi/kernels/funcs/eigen/sign.cc
index e71257f3f74aa..fe5868e5ef441 100644
--- a/paddle/phi/kernels/funcs/eigen/sign.cc
+++ b/paddle/phi/kernels/funcs/eigen/sign.cc
@@ -29,6 +29,7 @@ struct EigenSign<Eigen::DefaultDevice, T> {
   }
 };
 
+template struct EigenSign<Eigen::DefaultDevice, uint8_t>;
 template struct EigenSign<Eigen::DefaultDevice, int8_t>;
 template struct EigenSign<Eigen::DefaultDevice, int16_t>;
 template struct EigenSign<Eigen::DefaultDevice, int32_t>;
diff --git a/paddle/phi/kernels/funcs/eigen/sign.cu b/paddle/phi/kernels/funcs/eigen/sign.cu
index 58a4fe36232b6..88922bd86f2e2 100644
--- a/paddle/phi/kernels/funcs/eigen/sign.cu
+++ b/paddle/phi/kernels/funcs/eigen/sign.cu
@@ -29,6 +29,7 @@ struct EigenSign<Eigen::GpuDevice, T> {
   }
 };
 
+template struct EigenSign<Eigen::GpuDevice, uint8_t>;
 template struct EigenSign<Eigen::GpuDevice, int8_t>;
 template struct EigenSign<Eigen::GpuDevice, int16_t>;
 template struct EigenSign<Eigen::GpuDevice, int32_t>;
diff --git a/paddle/phi/kernels/gpu/sign_kernel.cu.cc b/paddle/phi/kernels/gpu/sign_kernel.cu.cc
index bbccc906a06e3..c160410ea6054 100644
--- a/paddle/phi/kernels/gpu/sign_kernel.cu.cc
+++ b/paddle/phi/kernels/gpu/sign_kernel.cu.cc
@@ -25,6 +25,7 @@ PD_REGISTER_KERNEL(sign,
                    GPU,
                    ALL_LAYOUT,
                    phi::SignKernel,
+                   uint8_t,
                    int8_t,
                    int16_t,
                    int32_t,
diff --git a/python/paddle/tensor/math.py b/python/paddle/tensor/math.py
index 85d1e1f8541c7..763d061872929 100644
--- a/python/paddle/tensor/math.py
+++ b/python/paddle/tensor/math.py
@@ -4596,7 +4596,7 @@ def sign(x, name=None):
     Returns sign of every element in `x`: 1 for positive, -1 for negative and 0 for zero.
 
     Args:
-        x (Tensor): The input tensor. The data type can be int8, int16, int32, int64, float16, float32 or float64.
+        x (Tensor): The input tensor. The data type can be uint8, int8, int16, int32, int64, float16, float32 or float64.
         name (str, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`.
 
     Returns:
@@ -4620,6 +4620,7 @@ def sign(x, name=None):
             x,
             'x',
             [
+                'uint8',
                 'int8',
                 'int16',
                 'int32',
@@ -4627,7 +4628,6 @@ def sign(x, name=None):
                 'float16',
                 'float32',
                 'float64',
-                'uint16',
             ],
             'sign',
         )
diff --git a/test/legacy_test/test_sign_op.py b/test/legacy_test/test_sign_op.py
index 92422a6fb9520..24a5983c279db 100644
--- a/test/legacy_test/test_sign_op.py
+++ b/test/legacy_test/test_sign_op.py
@@ -93,12 +93,16 @@ def test_dygraph(self):
             self.assertEqual((np_z == z_expected).all(), True)
 
     def test_static(self):
-        np_input2 = np.random.uniform(-10, 10, (12, 10)).astype("int16")
-        np_input3 = np.random.uniform(-10, 10, (12, 10)).astype("int32")
-        np_input4 = np.random.uniform(-10, 10, (12, 10)).astype("int64")
+        np_input1 = np.random.uniform(-10, 10, (12, 10)).astype("int8")
+        np_input2 = np.random.uniform(-10, 10, (12, 10)).astype("uint8")
+        np_input3 = np.random.uniform(-10, 10, (12, 10)).astype("int16")
+        np_input4 = np.random.uniform(-10, 10, (12, 10)).astype("int32")
+        np_input5 = np.random.uniform(-10, 10, (12, 10)).astype("int64")
+        np_out1 = np.sign(np_input1)
         np_out2 = np.sign(np_input2)
         np_out3 = np.sign(np_input3)
         np_out4 = np.sign(np_input4)
+        np_out5 = np.sign(np_input5)
 
         def run(place):
             with program_guard(Program(), Program()):
@@ -106,35 +110,47 @@ def run(place):
                 input1 = 12
                 self.assertRaises(TypeError, paddle.tensor.math.sign, input1)
                 # The result of sign_op must correct.
+                input1 = paddle.static.data(
+                    name='input1', shape=[12, 10], dtype="int8"
+                )
                 input2 = paddle.static.data(
-                    name='input2', shape=[12, 10], dtype="int16"
+                    name='input2', shape=[12, 10], dtype="uint8"
                 )
                 input3 = paddle.static.data(
-                    name='input3', shape=[12, 10], dtype="int32"
+                    name='input3', shape=[12, 10], dtype="int16"
                 )
                 input4 = paddle.static.data(
-                    name='input4', shape=[12, 10], dtype="int64"
+                    name='input4', shape=[12, 10], dtype="int32"
+                )
+                input5 = paddle.static.data(
+                    name='input5', shape=[12, 10], dtype="int64"
                 )
+                out1 = paddle.sign(input1)
                 out2 = paddle.sign(input2)
                 out3 = paddle.sign(input3)
                 out4 = paddle.sign(input4)
+                out5 = paddle.sign(input5)
                 exe = paddle.static.Executor(place)
-                res2, res3, res4 = exe.run(
+                res1, res2, res3, res4, res5 = exe.run(
                     paddle.static.default_main_program(),
                     feed={
+                        "input1": np_input1,
                         "input2": np_input2,
                         "input3": np_input3,
                         "input4": np_input4,
+                        "input5": np_input5,
                     },
-                    fetch_list=[out2, out3, out4],
+                    fetch_list=[out1, out2, out3, out4, out5],
                 )
+                self.assertEqual((res1 == np_out1).all(), True)
                 self.assertEqual((res2 == np_out2).all(), True)
                 self.assertEqual((res3 == np_out3).all(), True)
                 self.assertEqual((res4 == np_out4).all(), True)
-                input5 = paddle.static.data(
-                    name='input5', shape=[-1, 4], dtype="float16"
+                self.assertEqual((res5 == np_out5).all(), True)
+                input6 = paddle.static.data(
+                    name='input6', shape=[-1, 4], dtype="float16"
                 )
-                paddle.sign(input5)
+                paddle.sign(input6)
 
         for place in self.place:
             run(place)