Although pacemaker recalls are common, the optimal mechanism for risk assessment and triage of patients at risk for sudden loss of device system function is unknown. A retrospective chart review of 120 patients with factory proven failed devices was performed. Logistic regression analysis was used to determine clinical correlates of emergency room versus outpatient clinic presentation at time of device failure. Twenty-two patients (18%) presented to emergency and 98 (82%) to clinic. Sixty-three devices had no device output at the time of presentation. Multivariate logistic regression analysis revealed that antiarrhythmic drug use (odds ratio: 7.4, 95% CI: 2.0-28.0), atrioventricular nodal disease as an indication for pacing (odds ratio: 2.8, 95% CI: 1.2-3.0), and female gender (odds ratio: 2.2, 95% CI: 1.0-4.5) were the only significant correlates of emergency room presentations. Pacemaker dependency (escape heart rate < 40 beats/min) did not correlate with location of presentation even though no device output at the time of presentation was associated with emergency room presentation (odds ratio: 2.5, 95% CI: 1.1-5.8). Neither the presence of structural heart disease nor symptoms at the time of device implantation (syncope or presyncope) were correlated with location of presentation upon unexpected device failure. Although there were no deaths in the 120 failed devices studied, there were 26 deaths in the total group of 227 patients with recalled devices that could not be studied. Antiarrhythmic drug use, electrocardiographic pacing indication, and female gender may be more sensitive predictors of emergency room presentation and significant symptoms in the event of unanticipated pacemaker failure. The inability of any retrospective analysis to accurately assess mortality in the setting of pacemaker system failure underscores the need for prospective databases in recall situations.