Cardiac pacing is often considered in patients with recurrent syncope after repeated attempts to document the cause have failed. To assess the results of this tactic we reviewed the records of 104 patients who had received pacemakers for known or suspected bradycardia between September 1973 and March 1985. The patients were classified retrospectively into three groups: group 1 (31 patients with a mean age of 73 years) had unequivocal documentation of bradycardia during syncope, group 2 (42 patients with a mean age of 71 years) had electrocardiographic or electrophysiologic evidence of potential bradycardia but no documentation during spontaneous syncope, and group 3 (31 patients with a mean age of 69 years) had a history "suggestive of" bradycardia-related syncope but no other evidence to support the diagnosis. The rates of recurrence of syncope during follow-up were 6.3%, 7.3% and 32.2% in groups 1, 2 and 3 respectively (p less than 0.01). In group 3 recurrence was more probable in patients with loss of consciousness for more than 2 minutes than in those who were unconscious for 2 minutes or less (p less than 0.05). The results suggest that pacemaker implantation is justified for recurrent syncope after extensive attempts to document a spell have failed if abnormal diagnostic test results suggest bradycardia as a possible cause. Empirical pacing is less satisfactory in patients with normal results of evaluation but may arguably be justified when patients have recurrent syncope with injury.