Sudden heart arrest (HA) in the early phase after aorto coronary bypass surgery represents a serious event necessitating resuscitation, and for those who survive usually also an extra stay in the coronary care unit. Since such episodes of heart standstill may be related to conduction defects, a study was conducted to determine whether the duration of the QRS complex on the preoperative ECG is a marker for this morbid event. A cohort of 1011 consecutive patients operated on between 1982 and 1986 and followed to January 1st, 1993 were included in the study. Incidence of lethal or non-lethal HA during the first 4 weeks after surgery was considered as the primary endpoint and total mortality as the secondary endpoint. The incidence of HA was 40/1011 = 4%, with the majority of events (60%) being lethal. Independent risk factors of HA using the multivariate logistic model were previous coronary artery bypass surgery, presence of mitral regurgitation, left ventricular ejection fraction and the intraoperative cross-clamp time of aorta. Adjusting for the effect of confounder variables showed that the gradient effect of QRS complex duration on the endpoint HA was still present (p = 0.012). The duration of the QRS complex taken from the preoperative ECG had a gradient effect on the incidence of HA. With a baseline level of QRS <70 ms, the following odds ratios (OR) for HA were found: OR = 1.38 (95% CI 0.60-3.31) for QRS 70-80 ms; OR = 2.27 (95% CI 0.87-5.90) for QRS >90-120 ms; and OR = 3.38 (95% CI 1.06-11.50) for QRS > 120 ms, when adjusting for the risk factors. Cumulative survival at 5 years after surgery was 28+/-7.1% for patients experiencing HA versus 87+/-1.2% for patients free from this event. Our results underline the importance of the QRS complex duration as a preoperative marker for HA after aorta coronary bypass surgery, when adjusting for other risk factors. Although the one-year survival is poor for patients experiencing HA, there is no increase in mortality during the late follow-up.