The antifungal susceptibilities of 232 pathogenic blood stream Candida isolates collected during a recently completed trial comparing fluconazole (400 mg/day) with amphotericin B (0.5 mg/kg of body weight per day) as treatment for candidemia in the nonneutropenic patient were determined both by the National committee for Clinical Laboratory Standards M27-P macrobroth methodology and by a less cumbersome broth microdilution methodology. For amphotericin B, M27-P yielded a very narrow range of MICs (0.125 to 1 microgram/ml) and there were no susceptibility differences among species. For fluconazole, a broad range of MICs were seen (0.125 to > 64 micrograms/ml), with characteristic MICs seen for each species in the rank order Candida albicans < C. parapsilosis approximately equal to C. lusitaniae < C. glabrata approximately equal to C. krusei approximately equal to C. lipolytica. The MIC distribution for C. tropicalis was bimodal and could not be ranked. Both microdilution MICs were within one tube dilution of the M27-P MIC for > 90% of isolates with amphotericin B and for > or = 77% of isolates with fluconazole. For both methods, elevated MICs did not predict treatment failure. In the case of amphotericin B, the MIC range was too narrow to permit identification of resistant isolates. In the case of fluconazole, MICs for isolates associated with failure to clear the bloodstream consistently were equivalent to the median MIC for the given species. Successful courses of therapy were seen with four isolates from four patients despite MICs of > or = 32 micrograms/ml. As MICs obtained by M27-P and similar methods correlate with responsiveness to fluconazole therapy in animal models and in AIDS patients with oropharyngeal candidiasis, the lack of correlation in this setting suggests that the MICs for these isolates are at or below the relevant fluconazole breakpoint for this dose of fluconazole and patient setting and that host factors such as failure to exchange intravenous catheters were more important than MIC in predicting outcome.