OBJECTIVE--To examine the effects of pacing modes on the interaction between the left ventricle and arterial system in humans. DESIGN--The slope of the end systolic pressure-volume relation (end systolic elastance), effective arterial elastance, the ratio of effective arterial elastance to end systolic elastance, and mechanical energy efficiency were compared under different pacing modes (atrial, atrioventricular, and ventricular). PATIENTS--Nine male patients with sick sinus syndrome who had cardiac catheterisation for diagnosis and to see whether they needed a pacemaker. INTERVENTIONS--A conductance catheter with tip-manometer was inserted into the left ventricle to obtain pressure-volume loops, and two pacing catheters were inserted into the right atrium and into the right ventricle respectively. RESULTS--End systolic elastance was lower in atrioventricular pacing than in atrial pacing, but effective arterial elastance was not significantly different. End systolic elastance was lower in ventricular pacing than in atrioventricular pacing, and effective arterial elastance was higher in ventricular pacing than in atrioventricular pacing. Consequently the ratio of effective arterial elastance to end systolic elastance was lowest in atrial pacing and highest in ventricular pacing, and mechanical energy efficiency was highest in atrial pacing and lowest in ventricular pacing. CONCLUSIONS--Atrial contraction and synchronous ventricular contraction independently optimise ventriculoarterial coupling in terms of a transfer of energy. Thus atrial pacing gives the best ventriculo-arterial coupling among these pacing modes.