Abstract: An improved method is presented to get rid of disadvantages of conventional fuzzy-PID composite control after analyzing its control characteristics. The self-adaptive fuzzy-PID composite control is implemented via the fuzzy switching arithmetic based on trapezoidal membership function and the online self-tuning arithmetic with quantized and proportional factors based on variable universe and human simulating intelligence. With this improved method an intelligent temperature control system for PCR chip is performed to show its highly dynamic and stable traits, strong robustness and adaptability. Test results revealed that the improved controller is much better than conventional fuzzy-PID controller and simple fuzzy controller.