Objective: The objective is to test the concept of "pay for performance" for patients with non-small cell lung cancer. Methods: We constructed 53 benchmark performance standards (10 labeled "critical") and prospectively assessed the effect of adherence to these standards on morbidity and mortality for patients undergoing resection of non-small cell lung cancer. Results: Between January 1, 2007, and December 31, 2009, 778 patients with non-small cell lung cancer underwent thoracotomy by 1 surgeon. Ninety-seven percent of patients received all 26 of the "day of surgery" and "intraoperative" benchmarks, and those were the easiest to deliver. The 469 patients who had all 53 benchmarks delivered, compared with the 309 who did not, had a lower mortality (2.0% vs 2.3%) and morbidity (16% vs 44%; P < .001). The 693 patients who received all 10 "critical" benchmarks, compared with the 85 who did not, had a lower mortality (1.9% vs 4.7%) and morbidity (25% vs 41%; P = .003). Low household income and fewer than 2 people in the household were predictors of overall morbidity on univariate analysis. Conclusions: Most benchmarks, especially "day of surgery" and "intraoperative" ones, can be delivered in more than 97% of patients. The delivery of benchmarks reduces perioperative morbidity but not mortality. Socioeconomic factors are predictors of overall morbidity. Operative mortality is related to the "quality of the patient" and the "quality of the health care provider.".