Hydraulic screwdown system of cold strip mill is nonlinear, time-variant system and it is disturbed by many factors, so the turning quality of controller influences the accuracy of Automatic Gauge Control (AGC). In this paper an optimum design method based on particle swarm optimization (PSO) algorithm is presented for PID controller parameters optimization of hydraulic servo screwdown system of one cold strip mill. With PSO algorithm it is fast to calculate PID controller parameters through minimizing the fitness function and searching on the given controller parameter space. Simulation shows that the PID controller turned by PSO algorithm can achieve better control performance than Genetic Algorithm (GA) in AGC system of one single stand reversing cold rolling mill.