Modified thermal cycling conditions were explored in an effort to improve the reproducibility and resolving power of repetitive-element PCR (rep-PCR) fingerprinting. Assay performance was rigorously evaluated under standard and modified cycling conditions, using as a test set 12 strains putatively representing 12 serovars of Salmonella enterica. For all three fingerprint types (ERIC2, BOXA1R, and composite fingerprints), the use of extremely elevated annealing temperatures plus an initial 'touchdown' cycling routine yielded significant improvements in day-to-day reproducibility and discriminating power despite the somewhat sparser appearance of the fingerprints. Modified cycling conditions markedly reduced the variability of fingerprints between cyclers, allowing fingerprints from different cyclers to be analyzed together without the degradation of assay performance that occurred with between-cycler analyses under standard cycling conditions. With modified cycling, composite fingerprints exhibited the lowest reproducibility but the highest net discriminating power of the three fingerprint types. rep-PCR fingerprints led to the discovery of a serotyping error involving one of the 12 test strains. These data demonstrate that modified cycling regimens that incorporate elevated annealing temperatures (with or without an initial touchdown routine) may markedly improve the performance of rep-PCR fingerprinting as a bacterial typing tool.