To comparatively assess first-line treatment with fludarabine and 2 anthracycline-containing regimens, namely CAP (cyclophosphamide, doxorubicin plus prednisone) and ChOP (cyclophosphamide, vincristine, prednisone plus doxorubicin), in advanced stages of chronic lymphocytic leukemia (CLL), previously untreated patients with stage B or C CLL were randomly allocated to receive 6 monthly courses of either ChOP, CAP, or fludarabine (FAMP), stratified based on the Binet stages. End points were overall survival, treatment response, and tolerance. From June 1, 1990 to April 15, 1998, 938 patients (651 stage B and 287 stage C) were randomized in 73 centers. Compared to ChOP and FAMP, CAP induced lower overall remission rates (58.2%; ChOP, 71.5%; FAMP; 71.1%; P <.0001 for each), including lower clinical remission rates (CAP, 15.2%; ChOP, 29.6%; FAMP, 40.1%; P =.003). By contrast, median survival time did not differ significantly according to randomization (67, 70, and 69 months in the ChOP, CAP, and FAMP groups, respectively). Incidences of infections (< 5%) and autoimmune hemolytic anemia (< 2%) during the 6 courses were similar in the randomized groups, whereas fludarabine induced, compared to ChOP and CAP, more frequent protracted thrombocytopenia (P =.003) and less frequent nausea-vomiting (P =.003) and hair loss (P <.0001). For patients with stage B and C CLL first-line fludarabine and ChOP regimens both provided similar overall survival and close response rates, and better results than CAP. However, there was an increase in clinical remission rate and a trend toward a better tolerance of fludarabine over ChOP that may influence the choice between these regimens as front-line treatments in patients with CLL.