OBJECTIVE--To compare the different effects of right ventricular pacing and classic left bundle branch block on left ventricular function. DESIGN--Retrospective and prospective study of 48 patients by electrocardiography, and M mode, cross sectional, and Doppler echocardiography. SETTING--A tertiary cardiac referral centre. PATIENTS--48 patients (age range 21 to 89 years, 15 women), 24 with a VVI pacemaker implanted and 24 with classic left bundle branch block. Functional mitral regurgitation was present in all those with right ventricular pacing and 22 of those with left bundle branch block. RESULTS--Age, RR interval, and left ventricular size were similar in the two groups, as were conventional measurements of overall systolic function: shortening fraction and pre-ejection and aortic ejection times. In right ventricular pacing, however, QRS duration (p < 0.01) and electromechanical delay were much longer (p < 0.001), whereas the time intervals from onset of mitral regurgitation to aortic opening (contraction time) and from A 2 to the end of mitral regurgitation (relaxation time) were consistently shorter (p < 0.01) than corresponding values in patients with left bundle branch block. Reversed splitting of the second heart sound was much commoner in left bundle branch block (p < 0.02), and only these patients showed an early systolic ventricular septal contraction. Its onset followed the initial deflection of the QRS complex by 40(15) ms and preceded mitral regurgitation by a small but consistent interval of 10 ms (p < 0.01). The onset of posterior wall thickening was synchronous with the onset of mitral regurgitation in right ventricular pacing but much later (p < 0.01) in patients with left bundle branch block. The extent of incoordinate wall motion measure as relative dimension change during pre-ejection and isovolumic relaxation period was much greater (p < 0.01) in left bundle branch block. These major differences were not altered by left ventricular cavity size in either group, nor by the presence of previous left bundle branch block in patients who were subsequently paced. CONCLUSIONS--The left ventricle seems to be activated much more rapidly with right ventricular pacing than with left bundle branch block. This applies even when left bundle branch block is present before pacing. Electromechanical delay, contraction and relaxation times, and extent of incoordinate ventricular wall motion differ strikingly between the two conditions. The use of right ventricular pacing as an experimental model of left bundle branch block in humans must be re-examined.