Abstract: Soybean diseases are the important factors for restricting high-yield, high-quality, and high efficiency of sustainable agriculture, so soybean diseases should been diagnosed timely and accurately for intelligent agriculture. But the widely used traditional visual diagnostic methods could not meet the agricultural producers’ request of timely, because the high degree of subjective factors made it time-consuming and inaccurate. So it is very important to develop the intelligent disease diagnosis system. In this paper, we present a diagnostic method of soybean diseases based on uncertain reasoning of evidence credibility, which is combined evidence theory and uncertainty pass algorithm of credibility to overcome the complexity and uncertainty of the characteristic conditions for soybean disease expert system. In this paper, soybean diseases were set for study objects, and the automatic diagnosis model of soybean has been established through expression of credibility of knowledge for diseases of evidence and the use of pass algorithm for uncertainty reasoning rules, and the purpose of whose was to open up a new way for soybean disease remote system with efficiency and good robustness. There are two main stages of the proposed method. Firstly, the period, location and symptoms of soybean diseases characteristics and the prior knowledge of expert system are comprised as inferential credibility weights. Secondly, achieve the remote automatic diagnosis and decision-making for soybean disease used the proposed Uncertain Reasoning of Evidence Credibility method. The experiment carried out in reclamation of HeiLongJiang Province through dealing with actual data shows that the proposed method achieves a high accuracy of 87.62%, and has good adaptability, practicality as well as popularization. This method is a new efficient way for remote intelligent diagnostics and scientific management of plant diseases. In the later application, through uncertainty reasoning method and control strategy, the crop diseases intelligent diagnosis model with niche targeting can be set up, which according to many factors such as crop disease characteristics of credibility and priori knowledge of agricultural experts as well as characteristics of root, stipe, flower and fruit.