The Concorde trial shows an increase in CD4+ lymphocytes, but not a higher survival, in AZT treated asymptomatic patients. Our murine bone-marrow experiments show that in chronic AZT treatments a large increase in the lymphoids/erythroids compartment is due to host cytotoxicity. We put forward a mathematical formula for predicting cytotoxicity of given drug protocols, which can be used for modulating the schedule so as to increase its efficacy while maintaining its toxicity low. Our study suggests that in chronic treatments a large, single, daily dose will be less toxic than the same dose divided into several daily dosings.