Twenty-five patients seen consecutively at an HIV outpatient clinic who had clinical evidence of oropharyngeal candidiasis and two or more oral swabs positive for yeasts on culture were studied retrospectively. For each of the 65 isolates susceptibility to fluconazole was evaluated by the disk diffusion test and determination of the minimal inhibitory concentration (MIC). A correlation was sought between clinical resistance and in vitro susceptibility data. Seven patients were non-responders and 19 were responders (one patient figuring in both groups). Significant differences were observed between the two groups with respect to the median interval after the diagnosis of AIDS (27 months in non-responders and 2 months in responders; p = 0.001), the median CD4+ cell count (6 and 21 cells/mm3 respectively; p = 0.005) and the median number of previous episodes of oropharyngeal candidiasis treated with fluconazole (13 and 2 episodes respectively; p = 0.001). Candida albicans was identified in 64 of 65 cultures. The correlation between MIC values and diameters of inhibition was good (r = 0.85; p < 0.001). The degree of in vitro susceptibility of the isolates to fluconazole showed a significant difference between non-responders and responders (mean inhibition diameters 13 and 36 mm respectively; p < 0.001) with a tentative cut-off value of 25 mm. An advanced stage of HIV infection and previous exposure to fluconazole could be risk factors for the development of fluconazole-resistant oropharyngeal candidiasis. Candida albicans strains with decreased in vitro susceptibility to fluconazole were responsible for the clinical resistance which could be predicted by a simple disk diffusion test.