In order to improve the efficiency of cloud service selection and to guarantee trustworthy, available and reliable cloud service, a new model of cloud service selection based on trust trend was proposed. Base on this model, we firstly calculated the determined trust value in two parts: trust value and trust trend value (TTV). Trust value was calculated with Bayes theorem. Trust trend value was calculated with least squares linear regression. Trust trend value aims to illustrate the trust trend of changes in a given period. Then, we obtained the objective QoS value in the QoS quantitative model of could services. At the same time, the measuring strategy of trust relationship among cloud services was designed based on information entropy. The experiment result shows that this method can reflect changes in trust cloud services, enhance the predictive ability and effectively improve the success rate of cloud service selection.