Concerned about the inadequacy of a centralized database and the importance of low morbidity and mortality on carotid endarterectomy efficacy, the South Carolina Vascular Surgical Society prospectively instituted a computer registry for carotid procedures performed by its members, to establish a statewide standard of practice. From January 1994 through December 1997, 23 of the 30 physician members voluntarily registered data on 1652 carotid operations at 14 hospitals into a central database. Blinded results were reviewed biannually. Complete data (1995-1997) were available for 1199 cases. The patients tended to be >64 years old (72%), male (62%), and white (93%). Carotid endarterectomy was the most frequently performed operation (90%). Perioperative complications (< or = 30 days) occurred in 173 patients (14.4%), including stroke (n = 19; 1.6%), death (n = 8; 0.7%), and stroke/death (n = 25; 2.0%). Although 23 surgeons (77% of the society) contributed some data, only 10 surgeons (33%) contributed complete data on >10 patients/year. Despite biannual efforts to boost participation, case entry remained stable (1994, 358; 1995, 347; 1996, 425; and 1997, 427), representing about one-third of the estimated carotid procedures performed in the state during that period. The cost of the registry was approximately $11,500. Audit of 8 surgeons revealed a >95 per cent match against the statewide discharge database and low error rate versus independent medical record review. This experience confirms that excellent outcomes after carotid endarterectomy are not limited to a few select centers and can be accomplished by adequately trained surgeons in a variety of institutional settings. Incomplete physician participation, however, inevitably raises questions about the utility of such efforts. Until volunteer registries induce full participation by heightening perceived physician benefit, their role will remain limited for future outcomes research.