Background. The purpose of this study was to determine whether use of the left internal mammary artery (LIMA) during coronary revascularization influences short-term morbidity in all patients undergoing revascularization, as well as in patients over the age of 75 years, female patients, and patients with diabetes. The study also explored variability in the utilization of LIMA grafts across an entire state. Methods. Using the Clinical Outcomes Assessment Program (COAP) of the state of Washington, procedural outcomes were compared for patients receiving and patients not receiving LIMA grafts as part of revascularization procedures from January 1, 1999 to December 31, 2000. Mortality and major complications were examined, both as unadjusted rates and after adjusting for baseline patient risk factors. Results. A total of 16 centers performed 8,797 nonemergent coronary artery revascularizations, including 81.7% with LIMA grafts. The use of a LIMA graft was associated with a significantly lower mortality (3.7% No LIMA vs 1.6% LIMA), as well as decreases in ventricular arrhythmias, need for postoperative dialysis, need for transfusions, ventilator dependence, and length of hospital stay. These trends were true for the population as a whole as well as for all subgroups analyzed, and they persisted after correcting for differences in comorbid conditions. In addition, there was wide variability in the use of LIMA grafts from center to center in the state. Conclusions. The use of LIMA grafts for coronary revascularization is associated with decreased mortality and morbidity. Despite these advantages, there is great variability in its application across the state of Washington. © 2003 by The Society of Thoracic Surgeons.