A rapid method of antimicrobial susceptibility testing has been developed, which uses a modified microdilution procedure and an inoculum of 10 7 bacteria per ml. Results are determined within 4 h with an indicator consisting of 2( p -iodophenyl)-3( p -nitrophenyl)-5-phenyltetrazolium chloride. The precipitation of a red formazan by bacteria uninhibited by antimicrobials is accelerated by the addition of phenazine methosulfate. Isolates are classified as resistant, indeterminate, or susceptible, based on growth in up to two antimicrobial concentrations which conform closely to concentrations which correlate with the millimeter breakpoints used in the Bauer-Kirby method. Results of testing 10 antimicrobial agents against 1,126 isolates were compared with results obtained when the Bauer-Kirby method and the agar dilution procedure were used as reference methods. Enterococci were excluded because of false resistance. Discrepancies were classified as very major (false susceptibility), major (false resistance), and minor (combinations of susceptibility or resistance with indeterminate results). The rapid method versus the agar dilution method yielded 2.3% very major, 0.7% major, and 2.9% minor discrepancies, for a total of 6.0%. Of 58 organism-antimicrobial agent combinations tested, 23 displayed 1% very major discrepancies between the rapid method and the agar dilution method. Six were not therapeutically important. The remainder involved Staphylococcus aureus, Staphylococcus epidermidis, Acinetobacter sp., and most organisms tested with chloramphenicol. It is suggested that adjustments in antibiotic concentrations and/or inoculum size may eliminate these discrepancies. The rapid method appeared economical when compared with Autobac 1 and the Bauer-Kirby procedure.