The search and use of bioactive compounds for different applications have been investigated, since ancient time. Virtual screening (VS) has emerged as alternative methodological approach to the Combinatory Chemistry and High-Throughput Screening (HTS) in identifying novel drug candidates. In VS, only compounds that are selected applying different computational tools to huge virtual libraries of compounds are further tested in vitro. However, the effective use of VS model applications have some challenges such as the inherent complexity of the ligand-receptor interactions as well as by other factors such as ligand and receptor multiple conformations and also ligand metabolic stabilities and toxicities. Altogether these difficulties are hardly overcome using only one computational tool. Therefore, in the literature, it has been suggested to apply a sequence of different filters, such as filters that select compounds through similarity, pharmacophore and docking. In this review, we describe the advantages, limitations and examples of recent successful applications of some of these filters, including drug-like properties, structural properties, 2D similarity, pharmacophore, shape and docking filters. Moreover, we present the main steps involved in the preparation of virtual libraries of compounds that can be used in the VS.