Tourism is an important activity for the economy of several places in the world. In this context, the use of information systems might bring some benefits for tourists, for example, by helping people to select tourist attractions, such as restaurants, beaches, and museums. In this paper, we perform two evalu- ations of 22 recommendation algorithms provided by existing recommendation system libraries, aiming to identify efficient algorithms in terms of prediction accuracy. In the first evaluation, we compare the algorithms by measuring different metrics, such as RMSE, MAE, precision, recall, and F1 Score. In the second evaluation, we compare the recommendation algorithms based on the answers of 172 people that participated in our survey by evaluating different kinds of tourist attractions. The results of our study show that some recommendation algorithms remain on the top list with regards to efficiency on both studies, such as the SVD++, Baseline Only, and KNN Z Score with Pearson Baseline Similarity. Others are efficient in the first evaluation, or for some metrics, but not in the second study, for example, or the other way around. The results of our study are useful for people that are creating solutions in the tourism domain.