A clinically available model of an activity-sensing, rate-responsive pacemaker (Activitrax, Medtronic) utilizes body vibration during exercise as an indicator of the need for a rate increase. Although having the advantage of rapid onset of rate response, this system lacks specificity and the rate response does not closely correlate with the level of exertion. In addition, this pacemaker is susceptible to the effects of extraneous vibration. In this study involving 20 normal subjects fitted with an external Activitrax pacemaker, the rate responses to a variety of exercises were studied and were compared with the corresponding sinus rates. The vibration generated at the level of the pacemaker was also measured by accelerometers in three axes. Only a fair correlation (r = 0.51) was achieved between the pacemaker rate and the sinus rate. The total root mean square value of acceleration in either the anteroposterior or the vertical axes was found to have a better correlation (r = 0.8). As the main accelerations during physical activities were in the lower frequency range (0.1-4 Hz), a low-pass filter was used to reduce the influence of extraneous vibration. Selective sensing of the acceleration level may be usefully implemented in an algorithm for activity pacing.