Congestive heart failure due to advanced coronary artery disease or dilated cardiomyopathy is often associated with intraventricular conduction delays. Electrical resynchronization is an evolving method to improve clinical and functional status. To evaluate whether pacing-induced changes in the electrocardiogram are related to hemodynamic changes, we analyzed electrocardiograms of patients enrolled in the Pacing Therapies in Congestive Heart Failure trial. The study population consisted of 42 patients, New York Heart Association functional class III-IV with a baseline QRS complex of 175 +/- 32 msec and a PR interval of 196 +/- 33 msec. The mean left ventricular ejection fraction was 0.23. Using high-resolution computer scans, we measured QRS duration of intrinsic and paced electrocardiographs at different times during the study. Results of the electrocardiographic measurements were correlated with functional results. During the crossover period, 34 episodes of biventricular pacing, 27 episodes of left ventricular pacing, and 5 episodes of right ventricular pacing occurred, each at an individual optimized atrioventricular (AV) delay. The only significant difference was that right ventricular pacing increased the QRS width by 40 msec as compared with baseline or biventricular pacing. Functional benefit, as indicated by relative increase of peak oxygen uptake (VO2) compared with baseline, was significantly correlated with shortening of paced QRS width (correlation coefficient, r = 0.55; p <0.05). After 12-month follow-up of 28 patients, we saw a slight, nonsignificant decrease of intrinsic QRS width. With regard to the underlying disease, intrinsic QRS width at baseline and at 12 months was also not significantly different between patients with coronary artery disease and dilated cardiomyopathy. This study found that right ventricular pacing causes an increase in QRS duration in patients with left bundle-branch block, whereas in left ventricular and biventricular pacing, QRS width remains unchanged. Shortening of QRS width is correlated with a pronounced relative increase of peak VO2, and thus may become a noninvasive marker of clinical efficacy. There is no evidence of remodeling of the intrinsic electrocardiogram after 12 months of pacing.