OBJECTIVES: Studies examining costs following myocardial infarction (MI) have been limited by short follow-up, small sample sizes, restricted patient populations, or failure to include hospital, physician, or market characteristics. In a national sample with extensive supplemental data, we sought to identify the baseline factors predicting higher costs in the year following MI. METHODS: Elderly acute MI patients totaling 84,373 (90% white, 49% female) in the Cooperative Cardiovascular Project were linked to Medicare Part A claims, AHA Hospital Surveys, the CMS directory of physician specialties, and the CMS Hospital Wage Index File. Medicare charges were converted to costs using institutional cost to charge ratios. Associations with admission episode and one-year log-transformed costs were assessed by regression analysis with robust standard errors. RESULTS: Mean cost was $12,956 (median $8,833) for the admission episode and $19,597 (median $13,583) at one year. Patient characteristics accounted for 11% of the admission and one-year cost variation. Patient variables most highly associated with one-year costs included anterior MI, CHF, COPD, renal insufficiency, diabetes, and shock. Older age (age 80 years) was most strongly associated with lower costs. After adding hospital, physician, and market variables, the models explained 22% of admission and 17% of the one-year cost variation. Teaching hospitals and care by a cardiologist were both associated with higher one-year costs. While patient death was associated with lower admission episode costs, death after the initial episode was associated with higher one-year costs. CONCLUSION: In models examining the baseline predictors of one-year costs following MI, patient characteristics account for approximately twice the cost variation accounted for by other baseline variables.