This study was undertaken to determine the efficacy and toxicity of a shorter schedule of fludarabine administration (30 mg/m2 i.v. daily for 3 days every 4 weeks) in patients with previously treated chronic lymphocytic leukemia (CLL). Eighty patients with previously treated advanced (Rai III-IV) (54%) or progressive Rai stage 0-II (46%) were treated. The results of this trial were retrospectively compared with those of previous fludarabine trials using different schedules of administration. The complete response (CR), nodular CR and partial response rates were 10, 15 and 21%. The overall response rate (46%) was slightly lower than prior regimens using a 5-day schedule of fludarabine; however, this difference was not significant. Further evaluation by dual-parameter flow cytometry and immunoglobulin gene rearrangement analysis revealed that minimal residual disease was more common in a 3-day schedule. The overall incidence of infections per treatment course (14%) was significantly lower than that observed on the 5-day or weekly regimens (P < 0.001). The incidence of minor, atypical and viral infections were similar. There was no difference in survival in the various trials. In conclusion, a 3-day schedule of fludarabine in previously treated CLL patients was associated with a lower infection rate and similar survival to a 5-day schedule. These data support the use of a 3-day schedule of fludarabine as a single agent and in combination with other active agents.