This paper is to propose a new method of evaluating the efficiency of intentional precipitation, based on various randomized and non-randomized intentional precipitation modification experiments. This method will give a quantity evaluation of the efficiency of the intentional precipitation modification by using random samples from historical samples which exist in prominent ups and downs in time and space and finding no obvious difference between the samples and natural precipitation samples. The numerical simulation results show that the effect of the evaluation by this method responding to natural up-down precipitation is much smaller than other method. This method can analyze the ”signal” of intentional modification under the background of the natural up-down precipitation with more accuracy. This paper will also analyze the effective influence of the regional rainfall correlation, samples contrast and natural samples responding to evaluation results by numerical simulation experiments.