To construct and evaluate a decision analytic model of proposed management strategies for HIV-infected patients presenting with cerebral mass lesions, radiographically compatible with toxoplasmosis, lymphoma, or other etiologies, assuming knowledge of Toxoplasma antibody status in serum. Using decision analysis, we evaluated two management strategies, for patients found to be either Toxoplasma-seropositive or -negative, for whom an initial choice was made for early brain biopsy (EB) or for empiric therapy with delayed biopsy (ETDB) of non-responders. The outcome to be optimized was the percentage of patients alive at 12 months. Model variables included predictive value of toxoplasmosis serology, probabilities of treatment response and death within 14-21 days conditional on correct diagnosis, probability of operative death, probabilities of non-diagnostic brain biopsy conditional both on correct diagnosis and prior treatment. One and two-way sensitivity analyses, by Toxoplasma serostatus, led to the following conclusions (1) for Toxoplasma-seropositive patients, ETDB gives nearly equivalent outcomes to EB of all patients; (2) for Toxoplasma-seronegative patients, although both strategies have equivalent outcomes under baseline assumptions, EB is preferred if there are even small survival advantages for early versus delayed diagnosis of lymphoma or other conditions, or if risk of death within 14-21 days of ET exceeds 10% when correct diagnosis is not toxoplasmosis. Under plausible assumptions, Toxoplasma-seronegative patients will benefit from an early biopsy strategy.