diff --git a/paddle/fluid/API.spec b/paddle/fluid/API.spec
index c76489d183a5f1ccbccb4d36ec40ab200b6072c9..11ccef30f607061705284247b59acea930a4615e 100644
--- a/paddle/fluid/API.spec
+++ b/paddle/fluid/API.spec
@@ -56,7 +56,7 @@ paddle.fluid.io.save_persistables (ArgSpec(args=['executor', 'dirname', 'main_pr
 paddle.fluid.io.load_vars (ArgSpec(args=['executor', 'dirname', 'main_program', 'vars', 'predicate', 'filename'], varargs=None, keywords=None, defaults=(None, None, None, None)), ('document', '0a5308f496632ab1ec3ba1f1377e6f95'))
 paddle.fluid.io.load_params (ArgSpec(args=['executor', 'dirname', 'main_program', 'filename'], varargs=None, keywords=None, defaults=(None, None)), ('document', '41779819cef32f2246e83aebc5a002e2'))
 paddle.fluid.io.load_persistables (ArgSpec(args=['executor', 'dirname', 'main_program', 'filename'], varargs=None, keywords=None, defaults=(None, None)), ('document', '28df5bfe26ca7a077f91156abb0fe6d2'))
-paddle.fluid.io.save_inference_model (ArgSpec(args=['dirname', 'feeded_var_names', 'target_vars', 'executor', 'main_program', 'model_filename', 'params_filename', 'export_for_deployment'], varargs=None, keywords=None, defaults=(None, None, None, True)), ('document', '582d87b8df75a5a639a107db8ff86f9c'))
+paddle.fluid.io.save_inference_model (ArgSpec(args=['dirname', 'feeded_var_names', 'target_vars', 'executor', 'main_program', 'model_filename', 'params_filename', 'export_for_deployment'], varargs=None, keywords=None, defaults=(None, None, None, True)), ('document', '70f4f53f13572436ac72d1c8b5efeb9d'))
 paddle.fluid.io.load_inference_model (ArgSpec(args=['dirname', 'executor', 'model_filename', 'params_filename', 'pserver_endpoints'], varargs=None, keywords=None, defaults=(None, None, None)), ('document', '7a5255386075dac3c75b7058254fcdcb'))
 paddle.fluid.initializer.ConstantInitializer.__init__ (ArgSpec(args=['self', 'value', 'force_cpu'], varargs=None, keywords=None, defaults=(0.0, False)), ('document', '6adf97f83acf6453d4a6a4b1070f3754'))
 paddle.fluid.initializer.UniformInitializer.__init__ (ArgSpec(args=['self', 'low', 'high', 'seed'], varargs=None, keywords=None, defaults=(-1.0, 1.0, 0)), ('document', '6adf97f83acf6453d4a6a4b1070f3754'))
diff --git a/python/paddle/fluid/io.py b/python/paddle/fluid/io.py
index 1775159798414a98bede4a3db5b577fb5e47e611..326a84d82b5718dad898620a6d9e0490f7519448 100644
--- a/python/paddle/fluid/io.py
+++ b/python/paddle/fluid/io.py
@@ -896,7 +896,7 @@ def save_inference_model(dirname,
                                      True is supported.
 
     Returns:
-        None
+        target_var_name_list(list): The fetch variables' name list
 
     Raises:
         ValueError: If `feed_var_names` is not a list of basestring.
@@ -949,11 +949,13 @@ def save_inference_model(dirname,
     # TODO(Superjomn) add an IR pass to remove 1-scale op.
     with program_guard(main_program):
         uniq_target_vars = []
-        for var in target_vars:
+        for i, var in enumerate(target_vars):
             if isinstance(var, Variable):
-                var1 = layers.scale(var, 1.)
-            uniq_target_vars.append(var1)
+                var = layers.scale(
+                    var, 1., name="save_infer_model/scale_{}".format(i))
+            uniq_target_vars.append(var)
         target_vars = uniq_target_vars
+    target_var_name_list = [var.name for var in target_vars]
 
     # when a pserver and a trainer running on the same machine, mkdir may conflict
     try:
@@ -1010,6 +1012,7 @@ def save_inference_model(dirname,
         params_filename = os.path.basename(params_filename)
 
     save_persistables(executor, dirname, main_program, params_filename)
+    return target_var_name_list
 
 
 def load_inference_model(dirname,