Chemical libraries or databases are collections of compounds which can be screened (virtually or experimentally) in order to discover drug candidates. These libraries are very variable in their content (description of structures, molecular descriptors, literature links...) and their size (number of compounds). Over the last decade, a large number of papers have been published on the subject. In this review, we summarize these studies by introducing different types of compound collections and reviewing the main kinds of software used to manipulate them. We present the descriptors which have a fundamental role in the characterisation of the molecules, and describe how they are used to define the molecular filters applied before screening, in order to obtain both a representation of chemical spaces and selections of subsets by diversity or similarity.