From July 1983 to January 1991 a total of 622 patients were randomized (585 eligible) to compare the effects of hydroxyurea, interferon alpha (IFN), and busulfan on the duration of chronic phase, and survival. Further goals included the determination of prognostic parameters. 598 CML patients were documented and 575 evaluable. The Ph-status was known for 547 patients. 89.4% of the patients were Ph-positive (+). 11% had additional chromosome aberrations. The median survival of Ph+ patients by now is 4.2 years, that of Ph-patients 1.4 years. Ph-negative patients are older, tend to have lower cell counts and, as a group are more ill at diagnosis. A survival difference of about one year is expected between busulfan and hydroxyurea treated patients. Prospectively evaluated age, organomegaly related symptoms, Karnofsky index, extramedullary manifestations, number of erythroblasts and percent of circulating blasts proved to be of prognostic significance. A prognostic score (score 1) was determined which was superior to Sokal's score in the study population. 164 patients were randomized to receive IFN. In 54 patients (33%) IFN had to be terminated because of adverse effects, therapy resistance or other reasons. Clinically relevant neutralizing antibodies were detected in 9 cases. Most frequent adverse events were flu-like symptoms in 74%, gastrointestinal symptoms in 52%, and neurologic-psychiatric symptoms in 30% of patients. Reduction of the Ph-chromosome was observed in 13% of evaluable patients (10 of 75). In 4 patients complete cytogenetic remissions were observed, in three of these ongoing. Cytogenetic responders have a survival advantage. Interferon treated Philadelphia-negative CML patients have no survival disadvantage. The study is expected to allow statements as to the advantages or disadvantages of the use of busulfan, hydroxyurea and IFN in the treatment of CML as well as to the reliability of prognostic markers.