A typical grey prediction control system generally consists of a proportional-integral-derivative (PID) controller and a grey predictor implemented by GM(1,1). This study attempts to optimize such a grey prediction control system by use of differential evolution algorithm, where the parameters to be optimized are the coefficients of PID controller and the background value of grey model. In the proposed differential evolution-based optimization strategy, the optimal grey prediction control system is evaluated in terms of the maximal overshoot, rising time, steady-state error, and settling time. Simulation results on two single-variable plants demonstrate that differential evolution/best/1 performs the best among four differential evolution variants.