Background. Reduction of restenosis and reinterventions was recently reported with percutaneous interventions (PCI), including drug-eluting stents (Cypher; Cordis, Miami Lakes, FL). This study compares results of multivessel Cypher stenting with those of bilateral internal thoracic artery (BITA) grafting. Methods. From January 2002 to June 2004, 768 consecutive patients underwent multivessel myocardial revascularization; 138 by PCI including Cyphers and 630 by BITA. After matching for age, sex, ejection fraction, extent of coronary disease, and congestive heart failure, two groups (113 patients each) were used to compare the two revascularization modalities. Results. Both groups were similar; however, left main and intraaortic balloon were more prevalent in the BITA group. The number of coronary vessels treated per patient was higher in the BITA group (2.87 vs 2.22, p < 0.001). Follow-up ranged between 6 and 34 months. Thirty-day mortality was 0.9% in the BITA and zero in the PCI group (p = 0.32). There were no late deaths in the BITA and three (2.7%) in the Cypher group (p = 0.08). Angina returned in 28.3% of the Cypher and 12.4% of the BITA group, p = 0.003. A Cox proportional hazard model revealed assignment to the Cypher group to be the only predictor of angina recurrence (odds ratio 2.78, 95% confidence interval 1.46-2.56). There were 16 (14.2%) reinterventions in the Cypher group compared with six (5.3%) in the BITA group. One-year reintervention-free survival (Kaplan-Meier) of the BITA was 96% compared with 86.6% in the Cypher group (p = 0.005, log-rank test). Conclusions. Despite improved results of PCI with Cyphers, midterm clinical outcome of multivessel patients treated with BITA is still better.