Every adult acute nonlymphocytic leukemia patient in Rochester, New York seen from January 1975 to January 1982 was studied. Fifty percent of the patients did not receive combination chemotherapy. Among those who did, there was a significant selection bias toward placing patients with better prognostic features on protocol. Protocol patients were also treated with higher doses of chemotherapy than nonprotocol patients. However, these factors did not completely explain the significantly better complete response (CR) rate and survival among protocol patients. Eastern Cooperative Oncology Group (ECOG) participation remained an independent variable associated with a better outcome. An improvement in CR rate was seen during the 7-year period studied as compared to that seen between 1965 and 1974. The study provided evidence that the availability and use of ECOG protocols was a positive factor in the improvement of leukemia treatment in Rochester.