Recommendation systems apply knowledge discovery techniques to the problem of making personalized recommendation for information, products or services in the Internet. These works, especially collaborative filtering algorithms acquired relatively satisfactory results. They can millions of users easily search hundreds of millions of items. In tourism industry, potential customers may book hotel just once, which make it very important to recommend them suitable hotel. In this paper, user-based and item-based collaborative filtering methods have been proposed for hotel recommendation. Then we combine them to achieve better effectiveness. Experimental result demonstrates that our hybrid collaborative filtering algorithms outperform existing approaches.