diff --git a/doc/fluid/user_guides/howto/inference/image/image1.png b/doc/fluid/user_guides/howto/inference/image/image1.png
new file mode 100644
index 00000000000000..04e91da704b07f
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image1.png differ
diff --git a/doc/fluid/user_guides/howto/inference/image/image2.png b/doc/fluid/user_guides/howto/inference/image/image2.png
new file mode 100644
index 00000000000000..2d4ca01ebbffaa
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image2.png differ
diff --git a/doc/fluid/user_guides/howto/inference/image/image3.png b/doc/fluid/user_guides/howto/inference/image/image3.png
new file mode 100644
index 00000000000000..7eb8c16146175f
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image3.png differ
diff --git a/doc/fluid/user_guides/howto/inference/image/image4.png b/doc/fluid/user_guides/howto/inference/image/image4.png
new file mode 100644
index 00000000000000..34a0c21880e29a
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image4.png differ
diff --git a/doc/fluid/user_guides/howto/inference/image/image5.png b/doc/fluid/user_guides/howto/inference/image/image5.png
new file mode 100644
index 00000000000000..4aa85291858548
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image5.png differ
diff --git a/doc/fluid/user_guides/howto/inference/image/image6.png b/doc/fluid/user_guides/howto/inference/image/image6.png
new file mode 100644
index 00000000000000..499b1dc265d010
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image6.png differ
diff --git a/doc/fluid/user_guides/howto/inference/image/image7.png b/doc/fluid/user_guides/howto/inference/image/image7.png
new file mode 100644
index 00000000000000..a9f40af362a6e9
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image7.png differ
diff --git a/doc/fluid/user_guides/howto/inference/image/image8.png b/doc/fluid/user_guides/howto/inference/image/image8.png
new file mode 100644
index 00000000000000..6db078a7ae6efd
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image8.png differ
diff --git a/doc/fluid/user_guides/howto/inference/image/image9.png b/doc/fluid/user_guides/howto/inference/image/image9.png
new file mode 100644
index 00000000000000..f0dea70856a878
Binary files /dev/null and b/doc/fluid/user_guides/howto/inference/image/image9.png differ
diff --git a/doc/fluid/user_guides/howto/inference/index.rst b/doc/fluid/user_guides/howto/inference/index.rst
index b6452e09fb00ce..3d52e071ff3d69 100644
--- a/doc/fluid/user_guides/howto/inference/index.rst
+++ b/doc/fluid/user_guides/howto/inference/index.rst
@@ -10,3 +10,4 @@ PaddlePaddle Fluid 提供了 C++ API 来支持模型的部署上线
    build_and_install_lib_cn.rst
    native_infer.md
    paddle_tensorrt_infer.md
+   windows_inference.md
diff --git a/doc/fluid/user_guides/howto/inference/windows_cpp_inference.md b/doc/fluid/user_guides/howto/inference/windows_cpp_inference.md
new file mode 100644
index 00000000000000..f16c95111c0693
--- /dev/null
+++ b/doc/fluid/user_guides/howto/inference/windows_cpp_inference.md
@@ -0,0 +1,111 @@
+Windows环境模型预测使用说明
+===========================
+
+环境部署
+--------
+
+### 硬件环境
+
+测试环境硬件配置：
+
+| CPU   |      I7-8700K      |
+|----------|:-------------:|
+| 内存 |  16G |
+| 硬盘 |  1T hdd + 256G ssd |
+| 显卡 |  GTX1080 8G |
+
+测试环境操作系统使用win10 Version 18.03 版本。下载地址：
+
+### 环境配置步骤
+
+**一定要严格按照安装步骤顺序，否则会安装失败！**
+
+**安装vs2015**
+
+安装vs2015，安装选项中选择安装内容时勾选自定义，把关于c，c++，vc++的功能都安装上。下载地址：
+
+**安装CUDA8**
+
+需要去NVIDIA官网[https://www.geforce.cn/drivers](https://www.geforce.cn/drivers)
+下载显卡对应的驱动。推荐391版本
+
+![](image/image1.png)
+
+安装时需要勾选自定义，勾选安装全部。
+
+验证安装需要进入cmd中，输入nvcc -V查看。
+
+![](image/image2.png)
+
+如果有显卡安装驱动，也可以选择直接安装CUDA8.0，[https://developer.nvidia.com/cuda-80-ga2-download-archive](https://developer.nvidia.com/cuda-80-ga2-download-archive)
+
+**安装CUDNN**
+
+安装CUDNN只需要将文件中CUDNN
+7下的文件复制到对应的CUDA安装目录下。文件名，cudnn-8.0-windows10-x64-v7.zip。这里提供了cudnn
+7
+64位的版本。需要其他版本可在[https://developer.nvidia.com/cudnn](https://developer.nvidia.com/cudnn)
+下载。
+
+预测demo使用
+------------
+
+解压Paddle，Release，fluid\_install\_dir压缩包。
+
+进入Paddle/paddle/fluid/inference/api/demo\_ci目录，新建build目录并进入，然后使用cmake生成vs2015的solution文件。
+指令为：
+```cmake
+cmake .. -G \"Visual Studio 14 2015 Win64\" -DWITH\_GPU=ON
+-DWITH\_MKL=OFF -DWITH\_STATIC\_LIB=ON -DCMAKE\_BUILD\_TYPE=Release
+-DDEMO\_NAME=simple\_on\_word2vec
+-DPADDLE\_LIB=D:\\to\_the\_paddle\_fluid.lib
+-DCUDA\_LIB=D:\\CUDA\\v8.0\\lib\\x64
+```
+
+注：
+
+-DDEMO\_NAME 是要编译的文件
+
+-DPADDLE\_LIB 是fluid\_install\_dir路径，例如
+-DPADDLE\_LIB=D:\\fluid\_install\_dir
+
+-DCUDA\_LIB 是CUDA安装目录对应的文件夹
+
+Cmake可以在官网进行下载，并添加到环境变量中。[[https://cmake.org/download/]{.underline}](https://cmake.org/download/)
+
+执行完毕后，build目录如图所示，打开 箭头指向的solution文件：
+
+![](image/image3.png)
+
+修改编译属性为/MT：
+
+![](image/image4.png)
+
+![](image/image5.png)
+
+编译生成选项改成Release。
+
+![](image/image6.png)
+
+![](image/image7.png)
+
+将提供的依赖包中，Release下的openblas和模型文件拷贝到编译生成的Release下。
+
+![](image/image8.png)
+
+通过cmd进到Release目录执行：
+
+  1.  开启GLOG
+
+  set GLOG\_v=3
+
+  2.  进行预测
+
+  simple\_on\_word2vec.exe \--dirname=.\\word2vec.inference.model
+
+![](image/image9.png)
+
+FAQ：
+
+路径中尽量不要包含空格，例如发现CUDA\_LIB路径是Program
+Files(x86)可能会出错。可以将CUDA拷贝到一个新位置（这里直接拷贝就行）