Testing of Cryptococcus neoformans for susceptibility to antifungal drugs by standard microtiter methods has not been shown to correlate with clinical outcomes. This report describes a modified quantitative broth macrodilution susceptibility method showing a correlation with both the patient's quantitative biological response in the cerebrospinal fluid (CSF) and the survival of 85 patients treated with amphotericin B (AMB). The Spearman rank correlation between the quantitative in vitro measure of susceptibility and the quantitative measure of the number of organisms in the patient's CSF was 0.37 (P < 0.01; 95% confidence interval [95% CI], 0.20, 0.60) for the first susceptibility test replicate and 0.46 (P < 0.001; 95% CI, 0.21, 0.62) for the second susceptibility test replicate. The median in vitro estimated response (defined as the fungal burden after AMB treatment) at 1.5 mg/liter AMB for patients alive at day 14 was 5 CFU (95% CI, 3, 8), compared to 57 CFU (95% CI, 4, 832) for those who died before day 14. These exploratory results suggest that patients whose isolates show a quantitative in vitro susceptibility response below 10 CFU/ml were more likely to survive beyond day 14.