The intersection controlled by traffic lights is a high incidence zone. Human error may cause serious traffic accidents if the driver drives the vehicle to turn at such an intersection. In order to effectively evaluate the reliability of human factors in driving steering, on the basis of the cognitive reliability and error analysis method (CREAM), the decision-making trial and evaluation laboratory (DEMATEL) method is combined as the common performance conditions (CPC) to give weight. The formula of the total weight factor in the human error rate is improved to quantitatively predict the human error rate. The rationality of the method is verified and a comparative analysis with the method before the improvement is made. The results show that the improved CREAM method is more in line with the actual situation. This study provides a reference for the future quantitative study of the driver's error rate, which is also of great significance for improving driving safety and reducing the accident rate.