A method has been developed for the rapid identification of optimal enzyme substrates from combinatorial libraries. This methodology was validated by screening a 361-member N-terminally formylated tripeptide library, f-XXR (X = 19 different amino acids), for optimal substrates of Escherichia coli peptide deformylase (PDF). The library was synthesized on a solid phase via the split-pool synthesis method. The N-terminal formyl group was added by treating the resin with a 1:1 (mol/mol) mixture of HCO2H and DCO2D in the presence of dicyclohexylcarbodiimide. In a mass spectrum, each member of the library produced a doublet peak (separated by 1.0063 Da). Limited treatment of this library with E. coli PDF resulted in the deformylation of those peptides that are the most efficient substrates of the enzyme. The deformylated products, due to loss of the mass-degenerate formyl group, each generated a singlet peak in the mass spectrum. Thus, the PDF product peaks were readily identified and sequenced via tandem mass spectrometry. The results showed that PDF strongly prefers a norleucine and, to a lesser extent, a phenylalanine as the N-terminal residue, whereas it has little selectivity at the penultimate position. This result is in excellent agreement with the literature data and therefore demonstrates the methodology as an effective approach to the identification of optimal enzyme substrates. This method should be generally applicable to other enzymes as well as synthetic catalysts. Copyright © 2001 John Wiley & Sons, Ltd.