OBJECTIVE--To determine whether an interaction between encainide or flecainide and intercurrent ischaemia could account for the observed increase in cardiac and sudden deaths in the study group in the Cardiac Arrhythmia Suppression Trial (CAST) I. DESIGN--CAST I was a randomised, double blind, placebo controlled study in which patients received the drug which suppressed at least 6 premature ventricular contractions per minute by 80% or episodes of non-sustained ventricular tachycardia by 90%. Arrhythmic sudden death or aborted sudden death were the study end points. Measured secondary end points included recurrent myocardial infarction, new or increasing angina pectoris, congestive heart failure, and syncope. The CAST I database was analysed to determine which of three end points occurred first--cardiac death or cardiac arrest, angina pectoris, or non-fatal recurrent infarction. They were regarded as mutually exclusive end points. The triad of cardiac or sudden arrhythmic death plus congestive heart failure and syncope was similarly analysed. RESULTS--It was assumed that recurrent non-fatal infarction and new or increasing angina pectoris were ischaemic in origin. The sum of these non-fatal ischaemic end points and sudden death were nearly identical in the placebo group (N = 129) and the treatment group (N = 131). The one year event rate in each group was 21%. However, the treatment group had a much greater fatality rate (55 v 17; P < 0.0001) than the placebo group. The same relation was found when the data were examined on the basis of drug exposure rather than intention to treat. The temporal and circadian events were similar in each group and were consistent with an ischaemic pattern. No such patterns emerged from analysis of the presumed non-ischaemic end points of congestive heart failure and syncope. CONCLUSIONS--These data suggest that the interaction between active ischaemia and treatment with encainide or flecainide may have been responsible for the increased mortality seen in the treatment group in CAST I. This conversion of a non-fatal to a fatal event emphasises the need for future antiarrhythmic drugs to be screened in ischaemic models.