Early studies conflict regarding improved patient outcomes with cardiologist-directed care for acute myocardial infarction (AMI). We sought to assess the magnitude and mechanism of the influence of physician specialty on inpatient mortality for AMI. Using data from the Pennsylvania Health Care Cost Containment Council and elsewhere, we developed age-stratified logistic regression models of inpatient mortality, utilizing a split sample strategy for model development and validation. Referral bias and physician caseload were explicitly addressed. We analyzed 30,351 admissions for AMI. In patients < 65 years old, the adjusted odds ratio (OR) for mortality with cardiologist care was 0.89 (95% confidence interval [CI] 0.640 to 1.24, p = 0.49) relative to generalist care. In patients > or = 65 years of age, the adjusted OR was 0.86 (95% CI 0.72 to 1.03, p = 0.10). Caseload was significantly higher among cardiologists and was inversely related to inpatient mortality. Mortality models with caseload but not physician designation or physician designation without caseload found each predictor statistically significant in the absence of the other (OR for cardiologist care 0.82, 95% CI 0.71 to 0.95, p = 0.007; OR for patients with low volume physicians relative to high volume 1.27, 95% CI 1.05 to 1.51, p = 0.014). Older patients of physicians with higher case loads had a lower risk adjusted inpatient mortality for AMI. This probably explains the trend toward better outcomes among patients of cardiologists rather than noncardiologists.