At present,the durability test specifications of whole vehicle road simulation bench generally copy the proving ground,which ignores the difference between loads of the two and results in unsatisfactory bench test results. So it is difficult to reproduce the durability test results of the proving ground. Based on the measured load data of proving ground and test bench. We studied the difference between the two. The equivalent transfer criterion of load from road condition to bench condition is formed by using the principle of equal damage,and the ‘proving ground-bench ’ load-damage equivalent relationship model was established. The Pareto optimal solution was obtained by using fast and elitist non-dominated sorting in genetic algorithms（ NSGA-Ⅱ） algorithm to solve the model. By optimizing the solution,the road condition cycle number of the road simulation test bench durability evaluation was obtained,and the test criterion of the whole vehicle bench durability was formulated. The bench test was carried out by the new specification. The comprehensive validation by damage and the cumulative distribution of load was carried out. The results showed that the criterion for evaluating the durability of the whole vehicle road simulation test bench obtained by the method was efficient and accurate,which can effectively reproduced the evaluation effect in the proving ground.