Setting insurance rate is a key link of designing crop insurance. Accurate and reasonable premium rate will help to achieve crop insurance "farmers can afford, insurance company can compensate, the government can subsidy".This paper used parameter distribution method (including Normal distribution and Weibull distribution, Logistic distribution, etc) and nonparametric kernel density distribution method to simulate the probability distribution of the loss rate of sugarcane production, and set out the pure rate under different distribution respectively. At the same time, we used experience rating method to set the sugarcane insurance rates. The Study shown that, the pure rate by the experience rating method was lowest；in the parametric methods, the pure rate by the Normal distribution was lowest, the Weibull distribution was highest; over all, the fitting results of the parametric and nonparametric method were close, the result of the nonparametric method slightly higher than the parameter method. The study also showed that, to a large extent. The rational pure rate depends on the reasonable yield loss distribution model.