This review summarizes recent developments in fragment-based drug-discovery methods with an emphasis on informatics and modeling requirements. Fragment-based methods have become established as a powerful approach in structure-based lead discovery. A number of successful projects have been announced recently, where fragments have had a central role in hit generation and lead optimization, leading to candidates being considered for clinical trials. Despite these successes, there are still many opportunities for new development, such as improving the structural diversity of fragment libraries, strategies for fragment evolution, and methods for predicting fragment binding modes.