MicroRNAs (miRNAs) are non-coding RNAs containing 19-26 nucleotides, and they directly regulate the translation of mRNAs by binding to them. MiRNAs participate in various physiological processes and are associated with the development of diseases, such as cancer. Therefore, understanding miRNAs regulation on targets is crucial for understanding the mechanisms of diseases and for obtaining a more suitable treatment. In animals, the base complementarity between miRNAs and the mRNA is imperfect, hindering the prediction of these targets. Thus, over the past 15 years, several computational tools have emerged for the prediction of miRNA targets in animals, generally with a focus on human expression data. Taking into account the wide range of prediction tools, a systematic review is presented here to analyze and classify these methods and features to enable the most appropriate choice according to the needs of each researcher. In this study, only articles whose methods met the inclusion and exclusion criteria established in the protocol were considered. The search was performed in November 2020, in two search engines PubMed and VHL Regional Portal. Among the initial 5315 journals found in the two searches, 78 articles were accepted, comprising 49 different tools analyzed and grouped by features and method similarities. As we limited our criteria to animals, all tools found in our search were suitable for human studies. The results demonstrated the evolution of prediction tools, including the most used features, such as alignment and thermodynamics, the methods used, as well as performance issues. It is possible to conclude that the currently available miRNA target prediction tools and methods can be aggregated with new features or other methods to improve accuracy.