A growing number of cloud services have emerged on the Internet. In the face of a wide variety of cloud services, how to recommend high quality and low price service meeting the users' requirements to the user accurately has become a focus in cloud service recommendation field. Currently, many services recommendation methods are often just based on the current service status without taking into account the performance trend of cloud service. For this reason, this paper proposes a services recommendation model based on performance prediction. The model uses improved artificial neural network based on conjugate gradient to predict the performance of cloud services. Factors such as performance and prices can be compared and calculated by using AHP (Analytic Hierarchy Process), and then the most suitable cloud service would be recommended to the users. The experimental results show that the prediction accuracy would be higher by using improved neural network predicting service performance method, and AHP can recommend the most suitable service to the user according to comprehensively considering the performance and price of services.