The invention provides a probabilistic load flow output random variable evaluation method based on kernel density estimation. The method comprises the steps that 1, kernel density estimation of a probability density function of a random variable is calculated; 2, the bandwidth of kernel density estimation is calculated; and 3, Gaussian kernel-based kernel density estimation is calculated on the basis of a diffusion equation to obtain a probability distribution function of the output random variable. The method has the advantage that a rapid diffusion-based kernel density estimation algorithm is achieved.