Cationic antimicrobial peptides are able to kill a broad variety of Gram-negative and Gram positive bacteria and thus are good candidates for a new generation of antibiotics to treat multidrug-resistant bacteria. Here we describe a high-throughput method to screen large numbers of peptides for improved antimicrobial activity. The method relies on peptide synthesis on a cellulose support and a Pseudomonas aeruginosa strain that constitutively expresses bacterial luciferase. A complete substitution library of 12-amino-acid peptides based on a linearized variant (RLARIVVIRVAR-NH(2)) of the bovine peptide bactenecin was screened and used to determine which substitutions at each position of the peptide chain improved activity. By combining the most favorable substitutions, we designed optimized 12-mer peptides showing broad spectrum activities with minimal inhibitory concentrations (MIC) as low as 0.5 microg/ml against Escherichia coli. Similarly, we generated an 8-mer substituted peptide that showed broad spectrum activity, with an MIC of 2 microg/ml, against E. coli and Staphylococcus aureus.