The purpose of this study was to investigate the validity of automatically stored exercise data from the elastic band sensor compared with those of a gold-standard stretch sensor during exercises commonly used for rehabilitation of the hip and knee. The design was a concurrent validity study. Participants performed 3 sets of 10 repetitions of 6 exercises with both sensors attached to the same elastic exercise band. These were knee extension, knee flexion, hip abduction and adduction, hip flexion, and hip external rotation. Agreement between methods was calculated for date, time of day, repetitions, total and single repetition, and contraction phase-specific time under tension (TUT). Files from the elastic band sensor contained identical dates, time of day, and number of repetitions for each exercise set compared with those for the gold standard. Total TUT and total single repetition TUT were highly correlated with the stretch sensor (r = 0.83-0.96) but lower for contraction phase-specific TUTs (r = 0.45-0.94). There were systematic differences between the methods ranging from 0.0 to 2.2 seconds (0.0-6.3%) for total TUT and total single repetition TUT, and between 0.0 and 3.3 seconds (0.0-33.3%) for contraction phase-specific TUTs. The elastic band sensor is a valid measure of date, time of day, number of repetitions and sets, total TUT, and total single repetition TUT during commonly used home-based strength training exercises. However, the elastic band sensor seems unable to validly measure TUT for specific contraction phases.