Artificial neural network (ANN) and fuzzy math were introduced to the design filed of CNC software for realizing the fault tolerance of CNC interpolation and improving the reliability of software. In addition, function aspects (velocity, acceleration, chord error, prediction accuracy, fault tolerance, real time ) from the experiment on non-uniform rational B-spline (NURBS) interpolator based on ANN were evaluated in detail. The experimental results show that the NURBS interpolation based on ANN can not only meet the requirements of the function aspects, but also realize the fault tolerance of CNC interpolation, which may provide a new strategy in the improvement of the reliability of CNC software. 