Abstract: Lower Bayes confidence limits of one-shot products with little present data are discussed, using three kinds of prior distributions: Beta prior distribution with k factor, double prior distribution based on Beta distribution and double prior distribution based on uniform distribution. Through analysis and study, it is concluded that the result using Beta prior distribution with k factor proves to be better than that of the WCF method, and that double prior distribution based on Beta distribution is better than double prior distribution based on uniform distribution.