OBJECTIVES: This study aimed at performing an economic analysis in an American setting of the use of Angiotensin Converting Enzyme (ACE) inhibitor trandolapril in postinfarction patients as a preventative agent of heart failure (HF) progression, based on the TRACE trial's individual data. METHODS: The TRACE study was a prospective placebo-controlled clinical trial designed to determine the long-term effect of trandolapril in postinfarction patients with left ventricular dysfunction. 1749 patients were followed from 1992 to 1995. Our analysis was incremental and was conducted from a Payer Perspective in a US setting. Unit costs were attached to the use of resources (mean costs per Diagnosis Related Groups for hospitalizations and Average Wholesale Prices for concomitant dugs). The primary effectiveness criterion was the progression of HF as defined in the clinical trial, i.e. death from HF or hospitalization for HF or HF necessitating open-label ACE inhibitor administration. Uncertainty surrounding the estimate of the CE ratio was taken into account through a non-parametric bootstrap analysis. RESULTS: Trandolapril cost amounted to US$550 and concomitant drugs were valued to US$2386 for trandolapril treated patients versus US$1975 for placebo treated patients. On the other hand, hospitalizations cost was US$6671 for trandolapril patients versus US$7979 for placebo patients. The total mean medical cost was slightly lower in the active therapy group, US$9607 versus US$9954, p = 0.5305. 14.04% of trandolapril patients versus 19.70% of placebo patients experienced a HF progression, p = 0.003. Among 5,000 bootstrap re-samples, ACE inhibitor therapy was cost saving in 66.7% of the cases and cost-effective in 33.3%. The cost per avoided HF progression reached US$8860 among the samples with both cost and effectiveness differentials positive. CONCLUSIONS: These results obtained in an American setting could be considered as highly cost-effective. From a methodological standpoint, it raises the issue of CE analysis when cost differentials are close to zero.