Ventricular fibrillation (VF) has been induced at implantable cardioverter defibrillator (ICD) implant to ensure reliable sensing, detection, and defibrillation. Despite its risks, the value was self-evident for early ICDs: failure of defibrillation was common, recipients had a high risk of ventricular tachycardia (VT) or VF, and the only therapy for rapid VT or VF was a shock. Today, failure of defibrillation is rare, the risk of VT/VF is lower in some recipients, antitachycardia pacing is applied for fast VT, and vulnerability testing permits assessment of defibrillation efficacy without inducing VF in most patients. This review reappraises ICD implant testing. At implant, defibrillation success is influenced by both predictable and unpredictable factors, including those related to the patient, ICD system, drugs, and complications. For left pectoral implants of high-output ICDs, the probability of passing a 10 J safety margin is approximately 95%, the probability that a maximum output shock will defibrillate is approximately 99%, and the incidence of system revision based on testing is < or = 5%. Bayes' Theorem predicts that implant testing identifies < or = 50% of patients at high risk for unsuccessful defibrillation. Most patients who fail implant criteria have false negative tests and may undergo unnecessary revision of their ICD systems. The first-shock success rate for spontaneous VT/VF ranges from 83% to 93%, lower than that for induced VF. Thus, shocks for spontaneous VT/VF fail for reasons that are not evaluated at implant. Whether system revision based on implant testing improves this success rate is unknown. The risks of implant testing include those related to VF and those related to shocks alone. The former may be due to circulatory arrest alone or the combination of circulatory arrest and shocks. Vulnerability testing reduces risks related to VF, but not those related to shocks. Mortality from implant testing probably is 0.1-0.2%. Overall, VF should be induced to assess sensing in approximately 5% of ICD recipients. Defibrillation or vulnerability testing is indicated in 20-40% of recipients who can be identified as having a higher-than-usual probability of an inadequate defibrillation safety margin based on patient-specific factors. However, implant testing is too risky in approximately 5% of recipients and may not be worth the risks in 10-30%. In 25-50% of ICD recipients, testing cannot be identified as either critical or contraindicated.