It is of great importance to useany prior information effectively and reasonably in the evaluation of small samples. Therefore, a new testability evaluation method based on mixed Beta prior distribution is presented, while considering both the credibility and the importance of prior information as well as the testability evaluation of complex equipment in small samples. The results show that, according to classical methods using small binomial samples, the lowerconfidence limits of product testability are conservative. Most of the measurements for the credibility of the prior information are based on data. The evaluation results are aggressive due to the missing sources of prior information. Thus, the conclusion is reasonable, and this method is promising for engineering applications.