The application of an informative, iterative library design strategy is presented for lead identification and optimization. The computational algorithm underlying informative design systematically uses data from both active and inactive compounds and maximizes the information gained from subsequent design-synthesis-screening cycles. Retrospective analysis of a released dataset of 17 550 compounds and corresponding cyclin-dependent kinase-2 activities showed that informative library design yields significant enrichments of active compounds and efficiently discovers novel chemotypes in comparison with commonly used diversity-similarity protocols.