The defibrillation threshold (DFT) and upper limit of vulnerability (ULV) were determined using step-down protocols in 50 patients who underwent implantable cardioverter-defibrillator placement or testing. The sensitivity and specificity of each ULV energy level was assessed for detecting an increased DFT, correlation of the DFT and ULV, and optimal shock timing for ULV determination. A ULV <10 or 11 J (failure to induce ventricular fibrillation with 10- to 11-J shocks) was 100% predictive of an acceptable DFT and may be sufficient to exclude unacceptable DFTs in 60% of implantable cardioverter-defibrillator recipients. All 4 shocks used to scan the peak of the T wave during ULV testing were necessary for accurate ULV determination.