Existing points of interest recommendation systems do not consider either spatial-temporal factors etc or the historic behavior of users. Though both the two kinds of models have their strength, they don ’t make evaluation for accuracy rate of POIs recommendation, especially recommendation Satisfaction Index (RSI) of users. In this paper, a POIs recommendation model based on context-aware is built by combining the two ideas and it makes evaluation for RSI. Experimental results showed that, compared with the existing models, the performance of the model has large improvement on the aspects of accuracy rate and RSI.