Comparison of MIC results obtained in different parts of the world is currently difficult because of variations in methods. In this study, cation-adjusted Mueller-Hinton broth, the NCCLS-recommended medium, was compared with Iso-Sensitest broth, which is widely used in Europe. Microbroth dilution testing, using the NCCLS procedure, was performed on 124 Gram-positive (staphylococci and enterococci) and Gram-negative (Enterobacteriaceae and Pseudomonas aeruginosa) isolates from the CDC reference set, with the only variable being the medium used. Twelve antimicrobial agents were tested: amoxycillin-clavulanic acid, ampicillin, ciprofloxacin, erythromycin, gentamicin, imipenem, levofloxacin, oxacillin, gemifloxacin, trimethoprim- sulphamethoxazole, tetracycline and vancomycin. Vancomycin, erythromycin and oxacillin were only evaluated for the Gram-positive organisms. Trimethoprim-sulphamethoxazole was only evaluated for a subset of Gram-negative organisms because of off-scale results. The 124 isolates were tested in one American and one UK laboratory with two batches of cation-adjusted Mueller-Hinton broth and two of Iso-Sensitest broth. A statistical evaluation of the data used a 24 fully specified factorial analysis to determine if there were significant differences in results owing to Gram reaction, site of testing and type and/or batch of broth. In addition, the cumulative results for each antimicrobial agent in each broth were plotted against the range of MIC dilutions tested. MICs of ciprofloxacin, levofloxacin, gemifloxacin, gentamicin and tetracycline were slightly higher (half a doubling dilution) with Iso-Sensitest broth than with Mueller-Hinton broth. MIC results for the other antimicrobial agents were equivalent. Essential and category agreement rates were comparable for all agents (88.4-100% and 88.2-99.0%, respectively).