The major complication during therapy of chronic lymphocytic leukemia (CLL) with the purine nucleotide analogue fludarabine is infection, which is also the main cause of morbidity and mortality in the disease. As the incidence of infectious episodes during therapy correlated with severity of neutropenia, stage of disease, and response to therapy, an effort was made to reduce therapy-related myelosuppression and improve response by altering the conventional therapy regimen. The protocol which yielded a response rate of 57% in previously treated patients with CLL consisted of five consecutive daily doses of 25-30 mg/m2 fludarabine given every three to four weeks. Based on observations from intracellular pharmacology studies it was hypothesized that repetitive single weekly doses of fludarabine would allow normal bone marrow cells to recover while maintaining cytotoxic levels in the leukemic cells. The cumulative four-week dose of the once-weekly regimen was approximately 80% of the original protocol. Eleven out of 46 evaluable patients (24%) responded to the therapy. Seven patients (15%) achieved a complete remission, and four (9%) a partial remission. While myelosuppression was reduced by about 30% compared with the original protocol, the incidence of febrile episodes was increased by 17%. Pretreatment serum IgG levels below the normal range correlated significantly with a high incidence of infectious episodes and with a short median survival time. These observations suggest that in addition to myelosuppressive therapy, disease related depressed immune function causes morbidity and mortality due to infections. The results further show that changes in the scheduling of the therapy regimen, associated with a slightly lower dose, resulted in reduced efficacy as measured by the response rate.