The invention provides a method for calculating the partial discharge fault probability of a cable. The method comprises the following steps: 1) carrying out partial discharge and temperature detection on a to-be-detected cable, and continuously measuring partial discharge capacity and temperature data of the cable in n delta t time intervals; 2) calculating the increment delta Xi (i = 1, 2, ...,n) of the micropore of the electric tree in the ith delta t time period, solving a nonlinear equation set, and obtaining an estimated value L2i (i = 1, 2, ..., n) of the length l of the electric treeof the to-be-measured cable; 3) repeating the step (2) for n times to obtain estimated values L21 to L2n of the length of the electrical tree in n time periods, and carrying out non-parametric statistics on the estimated values L21 to L2n to obtain a probability density distribution function hFD (1) of the length l of the electrical tree of the cable to be measured; 4) counting the electric branchlength of the cable when the cable is close to the fault to obtain the probability density distribution fFD (L) of the electric branch length of the fault cable; and 5) solving the partial dischargefault probability FFD of the cable by integrating the hFD (1) and the fFD (L). The method has the beneficial effect that the purpose of quantitatively evaluating the fault probability of the XLPE cable by utilizing the partial discharge signal and the temperature data is achieved.