To facilitate successive Point-of-Interests (POI) recommendation, the categories of POIs and the regions where POIs are located are seldom considered in existing models. In view of this, we extend a state-of-the-art model SPENT, named SPENT+, by taking the category and the region into considerations. In SPENT+, we formulate category- and region-aware check-in sequences, design the similarity trees to aggregate similar features, and finally establish the category latent vectors and region latent vectors, respectively. The above two latent vectors are aggregated as the category-region-aware latent vectors. Therefore, the category-region-latent vectors are sent to an LSTM together with conventional check-in sequences to improve successive POI recommendation. We conduct two real datasets, Gowalla and Foursquare, and compare with state-of-the-art methods in experiments. Results show that SPENT+ outperforms the baselines in terms of precision and recall.