Azole-resistant thrush has emerged as a problem in people who are infected with human immunodeficiency virus (HIV) or acquired immunodeficiency syndrome (AIDS), especially those who have low CD4 cell counts who have had a previous relapse of oral candidiasis, and in those who require long-term suppressive antifungal therapy. Because of the development of a standardized methodology for antifungal susceptibility testing and interpretive criteria for resistance testing, studies of the clinical predictive value of in vitro results are possible. In this study, 61% of organisms isolated from patients who were receiving azole therapy and who had clinically resistant thrush had minimal inhibitory concentration values that would classify the isolate as "resistant" or "susceptible dose dependent." In contrast, 86% of isolates from patients with thrush that was clinically responsive to an azole were classified in vitro as "susceptible" or "susceptible dose dependent." No resistant isolates were detected in samples obtained from asymptomatic control patients who were not exposed to azole drugs. Serum levels of azole and CD4 cell counts were also important parameters with regard to prediction of response. We conclude that in vivo and in vitro correlations compare favorably to studies of susceptibility testing in bacteria.