The prognosis of patients with CML has improved little in the past 50 years. The relatively benign chronic phase invariably deteriorates to a refractory and rapidly fatal terminal phase. This terminal stage has been found to have two major subtypes as defined by morphologic, cytochemical, immunologic, and enzymatic criteria--myeloblastoid and lymphoblastoid. Aggressive combination chemotherapy has achieved minimal improvement in survival once the terminal phase has begun, perhaps because only Ph1-positive stem cells remain to repopulate the marrow at this stage. Bone marrow transplantation has also been unsuccessful as therapy for the terminal phase, possibly because the patients are too debilitated to tolerate transplantation once the terminal phase has begun. Combination chemotherapy has been applied in an effort to eliminate the Ph1 chromosome-containing clone during the chronic phase. This goal has not yet been consistently achieved. Chemotherapy has also not been able to delay the onset of the terminal phase nor to prolong survival. Even in those patients in whom the Ph1 chromosome-containing clone has been eliminated, relapse to the chronic phase with return of the Ph1 chromosome has generally occurred within a brief period of time. Bone marrow transplantation during the chronic phase may hold the promise of true cure for CML, with permanent elimination of the malignant clone. However, the chronic phase can be unpredictably long and patients in the chronic phase often have few, if any symptoms. Therefore, there has been a reluctance to employ drastic therapy during the chronic phase. Techniques to predict the transformation to the terminal phase prior to overt morphologic or clinical conversion are now being developed. It may be possible in the future to attempt HLA-matched sibling donor bone marrow transplantation at the earliest signs of transformation from the chronic to the terminal phase. In this manner, optimal survival might be achieved by allowing patients to be maintained in the chronic phase for as long as possible prior to the initiation of aggressive therapy. Until this is routinely possible, continued research designed to improve the therapy of the terminal phase must be pursued. These attempts are likely to include the development and evaluation of new chemotherapeutic agents, novel methods of administration of existing drugs to better exploit their pharmacokinetics (for example, continuous infusion), and the utilization of newly described treatment approaches (such as the use of "differentiating" agents in an attempt to prevent progression to blastic transformation).(ABSTRACT TRUNCATED AT 400 WORDS)