According to the stochastic behavior of the continuous phase plate (CPP) surface shape distribution, the autocorrelation function and the correlation length are used to analyze its surface shape characteristics. A Gaussian random distribution function is used to derive the analytic relation between the correlation length and the far-field distribution of CPP. Numerical algorithm is used to calculate the variance and the energy usage rate of CPP's far-field distribution, and also the specific influence of correlation length on far-field distribution of CPP. The conclusion is that the smaller the correlation length, the better the homogeneity of far-field distribution is, and the closer to the target spot the shape of the focal spot will be, and at the same time the higher the energy usage rate will be.