We have correlated the in vitro results of testing the susceptibility of Cryptococcus neoformans to fluconazole with the clinical outcome after fluconazole maintenance therapy in patients with AIDS-associated cryptococcal disease. A total of 28 isolates of C. neoformans from 25 patients (24 AIDS patients) were tested. The MICs were determined by the broth microdilution technique by following the modified guidelines described in National Committee for Clinical Standards (NCCLS) document M27-A, e.g., use of yeast nitrogen base medium and a final inoculum of 10⁴ CFU/ml. The fluconazole MIC at which 50% of isolates are inhibited (MIC₅₀) and MIC₉₀, obtained spectrophotometrically after 48 h of incubation, were 4 and 16 μg/ml, respectively. Of the 25 patients studied, 4 died of active cryptococcal disease and 2 died of other causes. Therapeutic failure was observed in five patients who were infected with isolates for which fluconazole MICs were ≥16 μg/ml. Four of these patients had previously had oropharyngeal candidiasis (OPC); three had previously had episodes of cryptococcal infection, and all five treatment failure patients had high cryptococcal antigen titers in either serum or cerebrospinal fluid (titers, >1:4,000). Although 14 of the 18 patients who responded to fluconazole therapy had previously had OPC infections, they each had only a single episode of cryptococcal infection. It appears that the clinical outcome after fluconazole maintenance therapy may be better when the infecting C. neoformans strain is inhibited by lower concentrations of fluconazole for eradication (MICs, <16 μg/ml) than when the patients are infected with strains that require higher fluconazole concentrations (MICs, ≥16 μg/ml). These findings also suggest that the MICs determined by the modified NCCLS microdilution method can be potential predictors of the clinical response to fluconazole therapy and may aid in the identification of patients who will not respond to fluconazole therapy.