diff --git a/python/paddle/trainer/config_parser.py b/python/paddle/trainer/config_parser.py
index 9c765b32d34b353e1abb25c602f4523f11cc5464..2a6b6d5e2bf86e67ddad0499743bcc06260924f2 100644
--- a/python/paddle/trainer/config_parser.py
+++ b/python/paddle/trainer/config_parser.py
@@ -2366,9 +2366,9 @@ class BatchNormLayer(LayerBase):
                  inputs,
                  bias=True,
                  use_global_stats=True,
-                 mean_var_names=None,
                  moving_average_fraction=0.9,
                  batch_norm_type=None,
+                 mean_var_names=None,
                  **xargs):
         if inputs is None:
             inputs = []
diff --git a/python/paddle/trainer_config_helpers/layers.py b/python/paddle/trainer_config_helpers/layers.py
index 4749f38c3800d579e6ad6383a383fc23dfcf6b71..e1703c158a599fe14740cf89062e9792ce76dab2 100644
--- a/python/paddle/trainer_config_helpers/layers.py
+++ b/python/paddle/trainer_config_helpers/layers.py
@@ -2957,10 +2957,10 @@ def batch_norm_layer(input,
                      bias_attr=None,
                      param_attr=None,
                      layer_attr=None,
-                     mean_var_names=None,
                      batch_norm_type=None,
                      moving_average_fraction=0.9,
-                     use_global_stats=None):
+                     use_global_stats=None,
+                     mean_var_names=None):
     """
     Batch Normalization Layer. The notation of this layer as follow.
 
@@ -3015,8 +3015,6 @@ def batch_norm_layer(input,
     :type param_attr: ParameterAttribute
     :param layer_attr: Extra Layer Attribute.
     :type layer_attr: ExtraLayerAttribute
-    :param mean_var_names: [mean name, variance name]
-    :type mean_var_names: string list
     :param use_global_stats: whether use moving mean/variance statistics
                              during testing peroid. If None or True,
                              it will use moving mean/variance statistics during
@@ -3029,6 +3027,8 @@ def batch_norm_layer(input,
                                    :math:`runningMean = newMean*(1-factor)
                                    + runningMean*factor`
     :type moving_average_fraction: float.
+    :param mean_var_names: [mean name, variance name]
+    :type mean_var_names: string list
     :return: LayerOutput object.
     :rtype: LayerOutput
     """
@@ -3047,10 +3047,10 @@ def batch_norm_layer(input,
         active_type=act.name,
         type=LayerType.BATCH_NORM_LAYER,
         batch_norm_type=batch_norm_type,
-        mean_var_names=mean_var_names,
         bias=ParamAttr.to_bias(bias_attr),
         moving_average_fraction=moving_average_fraction,
         use_global_stats=use_global_stats,
+        mean_var_names=mean_var_names,
         **ExtraLayerAttribute.to_kwargs(layer_attr))
 
     return LayerOutput(