Using a mathematical procedure, we determine appropriate sampling rates for logging heart rate, at a variety of exercise intensities. The mathematical procedure involves correlating exercise and heart rate data to determine a dynamical mathematical model, from which the frequency response of the relationship between exercise intensity and heart rate can be determined. The sampling rate is then straightforwardly deduced by making appropriate measurements on the frequency response curve. We show how careful consideration needs to be given to the choice of dynamical model structure and the work regimen, so that consistent and convincing conclusions can be drawn. We demonstrate that the dynamics of the workrate/ heart-rate system are dependent on the nominal work/heart rate, but a 5-s sampling period, as used in many commercial heart rate monitors, appears to be adequate, especially when some averaging is performed before logging.