Sinus node recovery time was compared with the recovery time of a slow atrioventricular junctional rhythm in each of the same seven pentobarbital anesthetized dogs. Recovery time and the first five cardiac cycles were examined after pacing atria and ventricles for 20, 40, and 60 seconds at four or more pacing cycle lengths. Data relating recovery times and return to control conditions to prepacing cycle length, pacing cycle length, duration of pacing, site of pacing, and origin of rhythms were analyzed by covariance analysis. From the analyses, the relative contribution of the determinants are: the prepacing cycle length 73%, the site of pacing 3.5%, the pacing cycle length 2%, and the interaction of the site of pacing and pacing cycle length 1% for sinus node recovery time; and for slow atrioventricular junctional rhythm recovery time, the duration of pacing 40%, the interactions between the duration of pacing and the pacing cycle length 27%, and the prepacing cycle length 9%. A modified exponential decay model predicted 8 beats for return to prepacing conditions during sinus rhythm and 66-100 beats during atrioventricular junctional rhythm. We conclude that the single most important determinant of sinus node recovery time is the prepacing cycle length. Pacing cycle length and site of pacing have a significant but small influence on sinus node recovery time and duration of pacing, beyond 20 seconds, has no significant influence. In contrast, duration of pacing is the most important determinant of atrioventricular junctional recovery time. Another major determinant of slow atrioventricular junctional recovery time is the interactions between pacing cycle length and duration of pacing. Prepacing cycle length has a minor influence, and site of pacing has no influence, on slow atrioventricular junctional recovery time.