Purpose: To inform coverage decisions regarding drug-eluting coronary stents by assessing their cost-effectiveness for particular high-risk patient subgroups. Methods: The analysis compared the cost-effectiveness of the two drug eluting stents (DES) available in the UK (Cypher and Taxus) and bare-metal stents (BMS). A review was undertaken of randomised trials including one or both of the DES. Individual patient data were available for some of these trials. The model parameters extracted from the trials were further revascularisation rates, undertaken for clinical reasons, either percutaneous interventions or bypass surgery, during follow-up. Three subgroups were considered on the basis of their higher baseline risk of repeat revascularisation: small vessels, long lesions and diabetics; a fourth subgroup without any of these risk factors was also considered. The evidence synthesis was implemented as a Bayesian hierarchical logistic model. Given that not all the trials directly compared all the therapies of interest, the model derived the absolute treatment differences based on the available mixed comparisons. A probabilistic decision model was developed incorporating the prices of the stents and the cost and utility effects of subsequent revascularisations. The analysis assumed that the stents had no differential effect on mortality. Results: Evidence from 15 trials was incorporated into the synthesis, 5 of which supplied individual patient data. The probability of repeat percutaneous interventions was lowest with Cypher and highest with BMS. Both DES had lower probabilities of subsequent bypass surgery than BMS, and the probability for Taxus was slightly lower than for Cypher. Both forms of further revascularisation were higher in patients with long lesions, small vessels or diabetes. Cost-effectiveness results were sensitive to the stent prices. At current UK prices (£908 BMS, £1300 Taxus and £1341 Cypher), Taxus is dominated by Cypher in diabetic patients and subject to extended dominance in all other subgroups. Cypher has an incremental cost per QALY gained, compared to BMS, of £13,759, £23,086, £13,740 in patients with small vessels, long lesions and diabetes, respectively. In patients with no risk factors, this increases to £35,865. Conclusion: Given existing list prices for the alternative stents, Cypher is likely to be considered more cost-effective than BMS and Taxus in patients at high baseline risk of further revascularization. This conclusion may change following any changes in the relative stent prices.