Cross-correlation interference is caused by C/A code correlation properties. It is an inevitable problem for GPS systems. And it has been one of the problems affecting the sensitivity of GPS receivers. To solve the problem, a novel auto-correlation detection method based on multi-groups data comparison (MDC) is presented. In this method, the Doppler frequency and code phase within several groups of data are compared to find the auto-correlation peak. In contrast to the mean square ratio method and peak difference method, the proposed method can improve the detection probability of weak signals effectively in the presence of cross-correlation interference. The simulation results show that in the presence of one strong signal with -124 dBm power level, the method can achieve an 83% detection probability for the weak signal with -145 dBm power level. Therefore, it is useful for the acquisition of weak GPS signal.