Through analyzing the installation method of laser interferometer measurement aids and the planning of measurement path, a direct measurement method of the error data of the feature points in the whole space of the machine tool is proposed according to the principle of laser interferometry. On this basis, a spatial interpolation algorithm is used to calculating the precision of quasi-feature points based on the error data of 8 vertices (that is feature points) of the small cube in machine tool space.The error data of a small cube has been increased from 8 to 125，which provides a large amount of sample data for the full-space precision modeling of machine tool. Finally, after obtaining the full-space feature points and quasi-feature point accuracy data of the machine tool, a neural network prediction model of machine tool full-space precision is established.Example application verification shows that the prediction precision exceeds 94%.