Cancer is one of the major causes of death in human beings. While traditional cancer treatments kill cancerous cells, they negatively affect normal cells. In addition, the side effects and high medical costs of treatment prevent effective management of cancer. Nonetheless, anticancer peptides have gained popularity over the recent years as potential therapeutic agents that may complement traditional therapies. Compared to conventional wet-lab experiments, computation-based methods provide a promising platform for high-throughput identification of peptides that have anticancer activity. Therefore, this review summarizes the currently available databases for anticancer peptides/proteins. This is a survey of 22 recently published in-silico methods that aim to predict anticancer peptides accurately. More specifically, the article details the benchmark datasets, feature construction, feature selection, machine learning algorithms, assessment criteria, comparison of different methods, and publicly available predictors. We also compare the prediction performance of these predictors to the benchmark dataset. Finally, the study makes several recommendations concerning the future development of databases for anticancer peptides and methods that can be used to predict anticancer peptides.