diff --git a/CMakeLists.txt b/CMakeLists.txt
index d3379a663db4613e529cdba4ce22111765ff59cc..7a7b5860a122a853fc9ce1da6494fc039b38bc10 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -305,6 +305,9 @@ if(WITH_DOC)
 endif()
 
 if (ON_INFER)
-    message(WARNING "On inference mode, will take place some specific optimization.")
+    message(STATUS "On inference mode, will take place some specific optimization.")
     add_definitions(-DPADDLE_ON_INFERENCE)
+else()
+    #TODO(luotao), combine this warning with `make inference_lib_dist` command.
+    message(WARNING "On inference mode, will take place some specific optimization. Turn on the ON_INFER flag when building inference_lib only.")
 endif()
diff --git a/cmake/inference_lib.cmake b/cmake/inference_lib.cmake
index 1047b6f998a74e42114b9deab4f0e7ba1af36835..efdb093a7b28e19f3b2a774dd54f2e7f042e9ca7 100644
--- a/cmake/inference_lib.cmake
+++ b/cmake/inference_lib.cmake
@@ -14,9 +14,6 @@
 
 # make package for paddle fluid shared and static library
 function(copy TARGET)
-    if (NOT ON_INFER)
-      message(WARNING "Turn on the ON_INFER flag when building inference_lib only.")
-    endif()
     set(options "")
     set(oneValueArgs "")
     set(multiValueArgs SRCS DSTS DEPS)
diff --git a/paddle/fluid/inference/api/demo_ci/run.sh b/paddle/fluid/inference/api/demo_ci/run.sh
index 340e84d9312c20e2d10eb4c0a13066511d5d2eb5..1ac655bdbbf7c45bfdde2c5fa8026fab2c891903 100755
--- a/paddle/fluid/inference/api/demo_ci/run.sh
+++ b/paddle/fluid/inference/api/demo_ci/run.sh
@@ -60,8 +60,7 @@ for WITH_STATIC_LIB in ON OFF; do
     -DWITH_MKL=$TURN_ON_MKL \
     -DDEMO_NAME=simple_on_word2vec \
     -DWITH_GPU=$TEST_GPU_CPU \
-    -DWITH_STATIC_LIB=$WITH_STATIC_LIB \
-    -DON_INFER=ON
+    -DWITH_STATIC_LIB=$WITH_STATIC_LIB
   make -j
   word2vec_model=${PADDLE_ROOT}'/build/python/paddle/fluid/tests/book/word2vec.inference.model'
   if [ -d $word2vec_model ]; then
@@ -81,8 +80,7 @@ for WITH_STATIC_LIB in ON OFF; do
     -DWITH_MKL=$TURN_ON_MKL \
     -DDEMO_NAME=vis_demo \
     -DWITH_GPU=$TEST_GPU_CPU \
-    -DWITH_STATIC_LIB=$WITH_STATIC_LIB \
-    -DON_INFER=ON
+    -DWITH_STATIC_LIB=$WITH_STATIC_LIB
   make -j
   for use_gpu in $use_gpu_list; do
     for vis_demo_name in $vis_demo_list; do
@@ -108,8 +106,7 @@ for WITH_STATIC_LIB in ON OFF; do
       -DWITH_STATIC_LIB=$WITH_STATIC_LIB \
       -DUSE_TENSORRT=$USE_TENSORRT \
       -DTENSORRT_INCLUDE_DIR=$TENSORRT_INCLUDE_DIR \
-      -DTENSORRT_LIB_DIR=$TENSORRT_LIB_DIR \
-      -DON_INFER=ON
+      -DTENSORRT_LIB_DIR=$TENSORRT_LIB_DIR
     make -j
     ./trt_mobilenet_demo \
       --modeldir=$DATA_DIR/mobilenet/model \