The results of intraoperative and postoperative predischarge implantable cardioverter-defibrillator (ICD) testing of 211 consecutive patients, starting at 15 J and requiring two successful terminations of induced VT/VF with a relative defibrillation safety margin (DSM) of >10 J, were reviewed. The aim was to define the type of intraoperative response that would make postoperative predischarge testing unnecessary. The intraoperative responses were divided into three types: A, a DSM > or =10 J and an absolute energy level of < or =20 J; B, a DSM of > or =10 J and an absolute energy level of >20 J; and C, a DSM <10 J and an absolute energy level of >20 J. At operation, the responses to defibrillation were A, 88.6%; B, 7.1%; and C, 4.3%. Accepting an A response only would leave 11.4% of the patients for postoperative testing. The positive and negative predictive values for diagnosing a postoperative C response were 0.78 and 0.97, respectively. Similarly, the predictive values for diagnosing a postoperative B or C response were 0.71 and 0.97, respectively. The postoperative testing responses were A, 89.1%; B, 4.3%; and C, 6.6%. In summary, an intraoperative A response was sufficient to make a postoperative defibrillation testing unnecessary, while it was found that intraoperative B and C responders should undergo postoperative testing. Applying these criteria, approximately 90% of the patients could be discharged without any postoperative induction test.