Two adaptive signal processing algorithms are applied to the estimation of the relative delay between Blood Volume Pulse (BVP) signals collected by independent sensors on the subject's index finger. Both the LMS Adaptive Predictor and the Adaptive Delay System are evaluated for the delay estimation in synthetic test signals and on actual pairs of BVP signals. The short-term variations in the relative delay of the BVP signals within a data record is studied and proposed as a key factor determining the better performance of the Adaptive Delay System. Changes in the estimated relative delay between the BVP signals of subjects before and after exercise are shown. It is proposed that these changes are associated with the modifications introduced in the cardiovascular system of the subject by exercise and can, therefore, be used to monitor the level of exercise achieved by a subject at different points in an exercise session.