Ninety-four newly diagnosed patients with hematologic malignancies were monitored for compliance with oral medication and scheduled clinic appointments over a 6-month treatment period. They were assigned at entry either to a control condition or an intervention condition designed to improve compliance. Compliance with medication was assessed objectively with serial serum drug and metabolite levels, as well as with self-report indices obtained on a monthly basis. Medical records were abstracted to obtain data on the number of appointments kept, treatment given, disease characteristics, and survival. On univariate analyses using the log-rank test, five variables were found to be significantly related to survival. These included compliance with allopurinol (probably acting as a surrogate for self-medication with other chemotherapeutic agents) (P = .007), control versus educational program cohort (P less than .001), disease severity (P = .009), Karnofsky at diagnosis (P = .011), sex (P = .084), and clinic appointments kept (P = .043). In hierarchical proportional hazards models, the following variables were associated with decreased risk of death: disease severity (P less than .025), keeping over 60% of appointments (P less than .05), allopurinol/oxipurinol compliance (P less than .01), and educational program cohort (P less than .05). After controlling for all other variables, three variables were associated with increased survival: high disease severity (relative risk [RR] = 2.48), high compliance with allopurinol (RR = .45), and educational program cohort (RR = .39). We conclude that compliance with oral medication has a significant effect on patient survival. In addition, the use of special educational and supportive programs designed to improve patient compliance are associated with significant prolongation of patient survival due to, as well as independent of their effects on compliance.