Previous studies have suggested that the level of residual disease at the end of therapy predicts outcome in chronic lymphocytic leukemia (CLL). However, available methods for detecting CLL cells are either insensitive or not routinely applicable. A flow cytometric assay was developed that can differentiate CLL cells from normal B cells on the basis of their CD19/CD5/CD20/CD79b expression. The assay is rapid and can detect one CLL cell in 10(4) to 10(5) leukocytes in all patients. We have compared this assay to conventional assessment in 104 patients treated with CAMPATH-1H and/or autologous transplant. During CAMPATH-1H therapy, circulating CLL cells were rapidly depleted in responding patients, but remained detectable in nonresponders. Patients with more than 0.01 x 10(9)/L circulating CLL cells always had significant (> 5%) marrow disease, and blood monitoring could be used to time marrow assessments. In 25 out of 104 patients achieving complete remission by National Cancer Institute (NCI) criteria, the detection of residual bone marrow disease at more than 0.05% of leukocytes in 6 out of 25 patients predicted significantly poorer event-free (P =.0001) and overall survival (P =.007). CLL cells are detectable at a median of 15.8 months (range, 5.5-41.8) posttreatment in 9 out of 18 evaluable patients with less than 0.05% CLL cells at end of treatment. All patients with detectable disease have progressively increasing disease levels on follow-up. The use of sensitive techniques, such as the flow assay described here, allow accurate quantitation of disease levels and provide an accurate method for guiding therapy and predicting outcome. These results suggest that the eradication of detectable disease may lead to improved survival and should be tested in future studies.