diff --git a/CHANGELOG.md b/CHANGELOG.md
index a556778ca..4a29d87ff 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -74,6 +74,7 @@ To release a new version, please update the changelog as followed:
 - Tutorials:
   - `tutorial_tfslim` has been introduced to show how to use `SlimNetsLayer` (by @2wins in #560).
 - Test:
+  - `test_utils_predict.py` added to reproduce and fix issue #288 (by @2wins in #566)
   - `Layer_DeformableConvolution_Test` added to reproduce issue #572 with deformable convolution (by @DEKHTIARJonathan in #573)
   - `Array_Op_Alphas_Test` and `Array_Op_Alphas_Like_Test` added to test `tensorlayer/array_ops.py` file (by @DEKHTIARJonathan in #580)
 - CI Tool:
@@ -93,7 +94,8 @@ To release a new version, please update the changelog as followed:
 ### Fixed
 - Issue #498 - Deprecation Warning Fix in `tl.layers.RNNLayer` with `inspect` (by @DEKHTIARJonathan in #574)
 - Issue #498 - Deprecation Warning Fix in `tl.files` with truth value of an empty array is ambiguous (by @DEKHTIARJonathan in #575)
-- Issue #572 with deformable convolution fixed (by @DEKHTIARJonathan in #573)
+- Issue #565 related to `tl.utils.predict` fixed - `np.hstack` problem in which the results for multiple batches are stacked along `axis=1` (by @2wins in #566)
+- Issue #572 with `tl.layers.DeformableConv2d` fixed (by @DEKHTIARJonathan in #573)
 - Typo of the document of ElementwiseLambdaLayer (by @zsdonghao in #588)
 
 ### Security
diff --git a/tensorlayer/utils.py b/tensorlayer/utils.py
index e316ca0d7..d65d36396 100644
--- a/tensorlayer/utils.py
+++ b/tensorlayer/utils.py
@@ -320,7 +320,7 @@ def predict(sess, network, X, x, y_op, batch_size=None):
             if result is None:
                 result = result_a
             else:
-                result = np.hstack((result, result_a))  # TODO: https://github.com/tensorlayer/tensorlayer/issues/288
+                result = np.concatenate((result, result_a))
         if result is None:
             if len(X) % batch_size != 0:
                 dp_dict = dict_to_one(network.all_drop)
@@ -338,7 +338,7 @@ def predict(sess, network, X, x, y_op, batch_size=None):
                 }
                 feed_dict.update(dp_dict)
                 result_a = sess.run(y_op, feed_dict=feed_dict)
-                result = np.hstack((result, result_a))  # TODO: https://github.com/tensorlayer/tensorlayer/issues/288
+                result = np.concatenate((result, result_a))
         return result
 
 
diff --git a/tests/test_utils_predict.py b/tests/test_utils_predict.py
new file mode 100644
index 000000000..dce212588
--- /dev/null
+++ b/tests/test_utils_predict.py
@@ -0,0 +1,51 @@
+import unittest
+
+try:
+    from tests.unittests_helper import CustomTestCase
+except ImportError:
+    from unittests_helper import CustomTestCase
+
+import tensorflow as tf
+import tensorlayer as tl
+import numpy as np
+
+
+class Util_Predict_Test(CustomTestCase):
+
+    @classmethod
+    def setUpClass(cls):
+        cls.x1 = tf.placeholder(tf.float32, [None, 5, 5, 3])
+        cls.x2 = tf.placeholder(tf.float32, [8, 5, 5, 3])
+        cls.X1 = np.ones([127, 5, 5, 3])
+        cls.X2 = np.ones([7, 5, 5, 3])
+        cls.batch_size = 8
+
+    @classmethod
+    def tearDownClass(cls):
+        tf.reset_default_graph()
+
+    def test_case1(self):
+        with self.assertNotRaises(Exception):
+            with tf.Session() as sess:
+                n = tl.layers.InputLayer(self.x1)
+                y = n.outputs
+                y_op = tf.nn.softmax(y)
+                tl.utils.predict(sess, n, self.X1, self.x1, y_op, batch_size=self.batch_size)
+                sess.close()
+
+    def test_case2(self):
+        with self.assertRaises(Exception):
+            with tf.Session() as sess:
+                n = tl.layers.InputLayer(self.x2)
+                y = n.outputs
+                y_op = tf.nn.softmax(y)
+                tl.utils.predict(sess, n, self.X2, self.x2, y_op, batch_size=self.batch_size)
+                sess.close()
+
+
+if __name__ == '__main__':
+
+    # tf.logging.set_verbosity(tf.logging.INFO)
+    tf.logging.set_verbosity(tf.logging.DEBUG)
+
+    unittest.main()