The effects of applied voltage and phase of respiration on impedance of pathways used by implantable cardioverter-defibrillators were investigated. Patients were studied at implantation of cardioverter-defibrillators using epicardial (n = 12) or transvenous and subcutaneous (SQ) (n = 30) electrodes. Transvenous-SQ pathways were right ventricular cathode to SQ anode and coronary sinus cathode to SQ anode. Transvenous-transvenous pathways were right ventricle to coronary sinus and right ventricle to superior vena cava. Patients with nonthoracotomy electrode systems were studied at end-expiration and end-inspiration. Five shocks of 65 to 745 V (0.2 to 34 J) were given in random order in sinus rhythm. Over this range, end-expiratory impedance decreased monotonically for all pathways. This effect was greatest for transvenous-SQ pathways (13 +/- 3% to 17 +/- 4%, p < 0.001), intermediate for transvenous-transvenous pathways (5 +/- 4% to 8 +/- 5%, p < 0.001), and least for epicardial pathways (3 +/- 3%, p = 0.006). Paired data in inspiration and expiration showed that inspiration increased impedance in transvenous-SQ pathways (p < 0.001) but not in transvenous-transvenous pathways. Further, the effects of respiration and voltage on impedance in transvenous-SQ pathways were interactive (p < 0.001): Inspiration increased voltage-dependence of impedance. The magnitude of the inverse relationship between voltage and impedance depends on type of defibrillation pathway. The effect of respiration on impedance suggests that voltage-dependence of impedance is greatest in the lungs. These findings have potential relevance for intraoperative testing of cardioverter-defibrillators and selection of pathways for low-energy cardioversion.