Abstract: In current evidential trust models, there exist some deficiencies such as local trust degree is sensitive to the threshold and the exact meaning of recommendation trust degree for trust target provided by recommender are obscure to the trust source, so an evidential trust model is proposed with fuzzy adjustment method to solve these problems. On the one hand, the trust ratings are adjusted through applying fuzzy sets, which makes local trust degree changes with the change parameter gradual and can avoid the occurrence of the mutation effectively. On the other hand, the exact meaning of recommendation trust degree can be comprehended through constructing a adjust method for the recommendation trust degree, which improve the accuracy of the trust metric calculated from the recommendation trust values. The simulation results show that compared with existing evidential trust models, the trust model in this paper has a strong ability in anti-disturbance and anti-attack, which is applied to a variety of dynamic environments, and has more remarkable enhancements in the accuracy of trust measurement.