In current researches of performance evaluation with evidential reasoning (ER) rule, the referential points of evidence are generally set as a quantitative value or an interval value. However, these two types cannot handle fuzzy uncertainty and probabilistic uncertainty simultaneously, which decreases the evaluation accuracy. In this paper, a new ER rule-based performance evaluation model for aerospace relay with distributed referential points is proposed. Firstly, in order to effectively embed the expert knowledge and capture the performance information, the Gaussian distribution is used to describe the referential points. Secondly, a complete performance evaluation model is constituted by calculating and optimizing the evidence parameters. Finally, the sensitivity analysis is performed on the model to locate the weak component of aerospace relay. A case study of the JRC-7 M aerospace relay is carried out to validate the effectiveness of the proposed model.