Abstract Summary: Non-ribosomal peptide synthetases (NRPSs) are multi-modular enzymes, which biosynthesize many important peptide compounds produced by bacteria and fungi. Some studies have revealed that an individual domain within the NRPSs shows significant substrate selectivity. The discovery and characterization of non-ribosomal peptides are of great interest for the biotechnological industries. We have applied computational mining methods in order to build a database of NRPSs modules that bind to specific substrates. We have used this database to build a hidden Markov model predictor of substrates that bind to a given NRPS. Availability: The database and the predictor are freely available on an easy-to-use website at www.nrpssp.com. Contact: carlos.prieto@unileon.es Supplementary information: Supplementary data is available at Bioinformatics online.