Reproducibility of the microdilution checkerboard method for antibiotic synergy

We assessed the reproducibility of the microdilution checkerboard method for measuring antibiotic synergy. Five strains of Pseudomonas aeruginosa were tested with four antibiotic combinations by using 10 replicates each. Twenty-five percent of replicate sets gave discordant classification results (i.e., a 7:3 or worse split in categorization). Determination of the individual MICs of each antibiotic alone was excellent; all 10 replicates were within 1 twofold dilution for 95% of the 80 sets of 10 replicates. The microdilution checkerboard method either should not be used or should be used with at least five replicates per determination, with > or = 80% agreement among the replicates required for classification.