Recently, the National Committee for Clinical Laboratory Standards (NCCLS) subcommittee on antifungal susceptibility testing published a proposed standard for a broth macrodilution method for in vitro susceptibility testing of yeasts. The major reason for developing such a method is to predict the likely clinical and microbiological outcome of treatment. A number of studies of fluconazole have attempted to correlate in vitro minimum inhibitory concentrations (MICs) with in vivo efficacy in humans by means of a variety of methods, including the NCCLS proposed standard. These studies have clearly demonstrated that the ability to predict clinical outcome with MICs is dependent on the patient population studies. Thus, fluconazole MICs were correlated with clinical outcomes for patients with AIDS-associated oropharyngeal candidiasis. In contrast, fluconazole MICs did not correlate with clinical outcomes for patients with candidemia who did not have AIDS; these patients have numerous factors that can independently influence the response to antifungal therapy aside from the MIC. Data on patients with AIDS-associated cryptococcal meningitis from my laboratory demonstrated that the use of fluconazole MICs in conjunction with clinical variables, such as blood culture results, enhances the ability to predict treatment outcome. More investigations addressing the role of host factors in the vitro-in vivo correlation are necessary before antifungal testing can be useful in guiding therapy.