In order to improve the precision of similarity calculation and recommendation accuracy and reduce data sparseness, we propose a collaborative filtering recommendation algorithm based on item attribute values and user features. Firstly, based on the user preference for item attribute values, we calculate the rating distribution of item attribute values and rating expectations, and obtain the user-attribute value rating matrix. In the meantime, we use a data similarity measure method to find user characteristics neighbors and fill the sparse user-attribute value rating matrix, thus obtaining the preference set of the nearest-neighbors. Thirdly, we calculate the rating of the unrated attribute values, and sort the means of the rating of all item attribute values, thus obtaining a Top-N recommendation list for the target user. Experiment on the Movie Lens data set and Book Crossing data set show that the algorithm can better overcome the data sparsity problem and enhance recommendation accuracy.