Pharmacologic prevention of ventricular fibrillation (VF) and sudden cardiac death (SCD) is based on the assumption that the abnormal impulse that causes premature ventricular complexes (PVCs) may trigger VF and that antiarrhythmic drug therapy will thus inactivate the trigger. However, results available from secondary preventive trials of antiarrhythmic agents, including the Ghent-Rotterdam Aprindine (GRAP) study, have not established conclusively the value of antiarrhythmic therapy. A number of factors could explain this failure, e.g., PVCs and VF might not be causally related, thus, antiarrhythmic drugs might not prevent VF; the number of patients enrolled in investigations might have been too small to accommodate statistical proof of benefit; or the subjects might have been poorly selected. A serious design defect that could interfere with demonstrating a benefit for treatment might arise from the fact that all patients in the GRAP and similarly designed studies had to show complex ventricular arrhythmias at randomization and that antiarrhythmic treatment was usually continued regardless of its effect on the underlying arrhythmias. Given the variability of arrhythmias, it would be expected that arrhythmias would persist in only a fraction of patients randomized to placebo. However, arrhythmias also persisted, although to a lesser extent, in treated patients. If one considers that SCD risk may be reduced in the face of arrhythmia suppression but that it may be increased if the arrhythmia persists due to a toxic or proarrhythmic drug effect, any advantageous effects of successful antiarrhythmic drug therapy would be offset by the negative effects of unsuccessful arrhythmia suppression, which would distort the comparison with placebo. The design of the GRAP and similar studies precludes stratification of data to obviate this effect. To discern whether a beneficial effect exists for arrhythmia suppression, the trial design must allow for initial treatment with the subject antiarrhythmic agent to establish its antiarrhythmic effect in the patient population, after which patients should be randomized to placebo or continuation of active drug therapy. A suitable design for such a study is discussed.