To determine how many electropharmacologic drug trials should be performed to select therapy for patients with ventricular tachyarrhythmias, the outcome of 150 consecutive patients with inducible ventricular tachyarrhythmias undergoing serial electropharmacologic testing was examined. The probability of identifying predicted effective therapy (inductive of fewer than five ventricular responses with three ventricular extrastimuli at three pacing cycle lengths) and the probability of that therapy preventing sustained ventricular tachyarrhythmia recurrences were determined as a function of the number of preceding trials. The probability ( +/- SE) of identifying predicted effective therapy by the first trial (0.23 +/- 0.03) was significantly higher than that of the second (0.09 +/- 0.04), third (0.08 +/- 0.04) and fourth (0.05 +/- 0.04) trials (p = 0.001). No patient had predicted effective therapy identified by subsequent trials. The 2 year actuarial probability of freedom from sustained ventricular tachyarrhythmias on predicted effective therapy was higher for the first (0.79 +/- 0.08), second (0.73 +/- 0.13) and third (0.86 +/- 0.13) trials than for the fourth (0.33 +/- 0.27) trial (p = 0.02). Thus, the probability of selecting therapy with long-term efficacy was highest for the first trial (0.18), intermediate for the second (0.07) and third (0.07) trials and lowest for the fourth (0.02) and subsequent (0.00) trials. Accordingly, the electropharmacologic approach to therapy selection should be abandoned after three unsuccessful trials.