The merit of carotid endarterectomy for patients who previously have sustained a completed stroke remains controversial. Between January 1976 and December 1983, 118 stroke patients with mild to severe permanent neurologic deficits were evaluated. Fifty-nine patients were managed nonoperatively and 59 operatively. Both cohorts were similar in age and sex distribution, incidence of hypertension (69%), diabetes mellitus (25%), and cardiac disease (39%). In the long-term follow-up (medical cohort average was 44.1 +/- 5.0 months; surgical cohort average, 41.8 +/- 3.7 months) the overall survival rate was comparable, that is, there were nine medical deaths and eight surgical deaths. However, there was a significant difference in the development of new neurologic deficits. Twelve of the 59 unoperated patients had new neurologic deficits and three patients died at 12, 36, and 48 months as a result of a recurrent stroke. New neurologic deficits developed in only two of the 59 surgical patients and there were no stroke-related deaths. When the cumulative probability of remaining free from recurrent deficits was examined in the surviving patients at 6 years, all of the patients in the operated group remained free from recurrent deficits, whereas only 58% of the patients in the unoperated group were free of new neurologic deficits (p = 0.02). These data suggest that stroke patients with fixed mild to moderate neurologic deficits and with carotid lesions may be protected from recurrent neurologic complications by carotid endarterectomy.