In summary, it is clear that in vitro susceptibility testing can predict outcome in selected clinical situations. The clearest data are from the fluconazole-treated AIDS patients with oropharyngeal candidiasis. In this setting, the homogeneity of the underlying immune defect, combined with the ease of identification and monitoring of the infection, creates a near-perfect test situation. In more complex scenarios, such as the heterogeneous population of patients enrolled in a recent study of candidemia, no such clear-cut correlation was present. The importance of host factors in the correlation of the MIC with outcome cannot be overemphasized. Examples of these parameters include patient status (underlying disease, the presence of intravascular catheters, and CD4+ T-cell number), drug pharmacokinetics (absorption and distribution), patient compliance, and drug-drug interactions. Identification of relevant factors can substantially improve the degree of the MIC-outcome correlation and thus improve the clinical utility of in vitro testing. An important feature in this entire process is the role of standardized susceptibility testing procedures. While not without flaws, the proposed NCCLS reference method has been invaluable in allowing multiple investigators to contribute data that can be used to clarify the correlation between the fluconazole MIC and outcome. While the development of simplified second-generation methods is eagerly anticipated, the role of the reference method as a common touchstone is critical. Only by use of either the reference method itself or methods with a known relationship to the reference method can this broad collaborative process really proceed. Current work is focusing on defining interpretive breakpoints for fluconazole and Candida species, refinement of the in vitro procedures used to measure susceptibility to amphotericin B, ketoconazole, and itraconazole, and the acquisition of a broad base of data on the relationship between the MIC and outcome for these three drugs. Although considerable work remains to be done, the available data suggest that solutions to each of these problems are possible and that routine susceptibility testing of fungi will become meaningful for clinical decision making in the foreseeable future.