The aim of this study was to estimate the incremental cost effectiveness of replacing bare metal stents (BMS) by drug-eluting stents (DES) when using trial data and registry data. We developed a Markov model (model of cost effectiveness of coronary artery disease) in which 60-year-old patients started by undergoing percutaneous coronary intervention for acute or subacute coronary artery disease. The patients are followed until death or 100 years of age. Data on the occurrence of events (revascularization, acute myocardial infarction, and death) were based on Scandinavian registry data. Separate analyses were conducted with data on effectiveness based on randomized controlled trials and patient registries. On using trial data, it was found that sirolimus-eluting stents (SES) yield 0.003 greater life expectancy and $3300 lower costs than do BMS (dominant strategy). Paclitaxel-eluting stents (PES) yield 0.148 more life years than do SES at additional lifetime costs of $2800 ($21,400 per life year gained). On using registry data, the cost per life year gained was found to be $4900 when replacing BMS with DES. Probabilistic sensitivity analyses, on the other hand, indicate that PES only has a 50%-75% probability of being cost effective, regardless of the type of effectiveness data. DESs are cost effective with current willingness to pay for life year gains. Whether PES or SES is the most effective DES remains uncertain.