Fludarabine is the most active agent in the treatment of chronic lymphocytic leukemia (CLL). Despite this activity only a minority of patients treated with fludarabine achieve a complete response. We evaluated a new treatment program of sequential therapy with fludarabine followed by high-dose cyclophosphamide in previously untreated patients with CLL. This report details the results in 25 patients with previously untreated CLL. Patients received fludarabine (25 mg/m2/day x 5 days every 4 weeks for six cycles) as induction followed by consolidation with high-dose cyclophosphamide at one of three dose levels 1.5 g/m2, 2.25 g/m2, or 3 g/m2 administered every 2 weeks for three doses. High-dose cyclophosphamide was given with G-CSF support (5 microg/kg/day days 3-12). Complete response (CR) required a normal physical examination, normal CBC, a normal bone marrow evaluation including no residual lymphoid nodules on biopsy. A nodular response was defined as a complete response with the exception of an occasional residual nodule seen on bone marrow biopsy. Flow cytometric analysis for CD5:CD19 dual staining and kappa/lambda clonal excess was performed in all patients as a sensitive measure of minimal residual disease (MRD). Selected patients had patient/tumor-specific oligonucleotides generated that were subsequently used in a polymerase chain reaction as an extremely sensitive measure of MRD. There were no treatment-related deaths and no patient encountered unacceptable toxicity. After completion of this sequential regimen 76% (95% confidence interval: 59-93%) of patients had a major response: eight (32%) achieved a CR, four (16%) a nodular response, seven (28%) a PR, and six patients (24%) failed. Four patients withdrew from study during induction with fludarabine and did not receive at least one cycle of cyclophosphamide. Of the 21 patients who received consolidation with cyclophosphamide 10 (48%) had an improved quality of response when compared to that achieved with fludarabine. Two patients (8%) had no disease detectable by flow cytometry ('flow cytometric' CR) after six cycles of fludarabine. This improved to nine patients (36%) after high-dose cyclophosphamide. Following consolidation with high-dose cyclophosphamide three patients (12%) tested negative by PCR. All of these patients had morphologic evidence of residual disease after six cycles of fludarabine. Consolidation with high-dose cyclophosphamide increased the fraction of patients achieving a nodular response or CR three-fold (16% to 48%). This appears to be clinically relevant because with a median follow-up of 52 (range 34-78) months the projected 6-year survival for patients achieving a CR or NR is 91% compared to 41% for all others (P = 0.012). We conclude that sequential therapy with fludarabine followed by high-dose cyclophosphamide in previously untreated patients with CLL is safe and can improve the quality of response in a large proportion of patients compared to therapy with fludarabine alone.