SummaryTreatment with zidovudine has been standard therapy for patients with advanced HIV infection, but intolerance is common. Previously, management of intolerance has consisted of symptomatic therapy, dose interruption/discontinuation, and, when appropriate, transfusion. The availability of new antiretroviral agents such as didanosine as well as adjunctive recombinant hematopoietic growth factors makes additional strategies possible for the zidovudine-intolerant patient. Because all of these agents are costly, we evaluated the cost implications of these various strategies for the management of zidovudine-intolerant individuals within a population of persons with advanced HIV disease. We performed a decision analysis using iterative algorithmic models of 1 year of antiretroviral care under various strategies. The real costs providing antiretroviral therapy were estimated by deflating medical center charges by specific Medi-Cal (Medicaid) charge-to-payment ratios. Clinical data were extracted from the medical literature, product package inserts, investigator updates, and personal communications. Sensitivity analysis was used to test the effect of error in the estimation of parameters. The models predict that a strategy of dose interruption and transfusion for zidovudine intolerance will provide an average of 46 weeks of therapy per year to the average patient at a cost of $5,555/year of therapy provided (1991 U.S. dollars). The models predict that a strategy of adding hematopoietic growth factors to the regimen of appropriate patients would increase the average amount of therapy provided to the average patient by 3 weeks (6%) and the costs attributable to therapy by 77% to $9,805/year of therapy provided. Finally, they predict that a strategy of switching intolerant patients to didanosine would increase the average amount of therapy provided to the average patient by 5 weeks (9%) and decrease costs attributable to therapy by 13% to $4,844/year of therapy provided. Increases in overall cost for strategies using hematopoietic growth factors were driven by large drug costs, while savings for strategies using crossover to didanosine were driven by savings on toxicity management and monitoring. A strategy of adding hematopoietic growth factors to the regimen of appropriate zidovudine-intolerant patients in a population would result in a modest increase in the total amount of antiretroviral therapy provided to that population but would also greatly increase overall costs. Crossing such patients over to didanosine yields a greater increase in the amount of therapy provided to the population under care and a decrease in overall costs. This occurs because use of growth factors in the zidovudine-intolerant patients