Purpose: The benefits of exercise for older adults with knee osteoarthritis (OA) are undisputed. Given the chronic nature of OA, home exercise programs provide an economic and sustainable intervention option. However, for these programs to be beneficial, patient adherence is crucial. Self-reported adherence measures such as exercise diaries and simple self-rated scales are commonly used both clinically and in research, yet it has been acknowledged that these measures are vulnerable to reporting bias, particular over increasing periods of recall, and little evidence exists demonstrating their validity or reliability. The objectives of this study were threefold: (i) to examine concurrent validity of adherence over 12 weeks measured via an exercise diary compared to adherence simultaneously measured using a concealed accelerometer in a cuff weight, among a cohort of older adults with chronic knee pain undertaking a home strengthening program; (ii) to examine the validity of a self-rated adherence scale with increasing duration of recall from 2-12 weeks compared to accelerometer measured adherence; and (iii) to evaluate test-retest reliability of the self-rated adherence scale. Methods: 54 adults aged 45 years or older with self-reported painful knee OA completed a physiotherapist-prescribed quadriceps strengthening program using an ankle cuff weight. Over 12 weeks participants completed daily home exercise diaries (from which exercise volume was calculated by multiplying the number of exercise sessions recorded (possible range of 0-10 for each 2-week period) by the number of exercises performed in each session recorded (possible range of 1-4 exercises)) and at fortnightly appointments rated their agreement with the statement “I have been doing my exercises exactly as I was asked to by my physiotherapist” since Week 0 on an 11-point Numeric Rating Scale (0 = Strongly disagree to 10 = Strongly agree). A triaxial accelerometer and data logger were concealed (without the participant's knowledge) in the ankle cuff weight used for exercises. To enable evaluation of test-retest reliability of the NRS at Week 2 participants repeated the scale at home the day following their physiotherapy session. Data were analysed using descriptive statistics, Spearman rank-order correlations, the Wilcoxon matched-pairs signed-rank test and the Bland-Altman method to assess agreement. Intra-class correlation coefficients (ICC's) were used to assess test-retest reliability. Results: Adherence was significantly overestimated in exercise diaries over the 12 weeks compared to accelerometer recordings (Diary median = 220 exercises; Accelerometer median = 176; z = 6.2; P<0.00). A Bland-Altman plot indicated large between-participant variability in level of agreement between these measures. Neither self-reported measure met the optimal level of correlation (r>0.80). Adherence recorded in exercise diaries showed a moderate correlation with accelerometer data over 12 weeks (r = 0.52 (95%CI 0.26–0.69)). Correlations between self-rated adherence and accelerometer data over 2-12 weeks were little to fair with wide confidence intervals (r = 0.23 (95%CI 0.003–0.47) – r = 0.39 (95%CI 0.14–0.58)). No pattern of correlation was evident with increasing recall time periods. On the two occasions one day apart at Week 2, self-rated adherence neared, but did not meet minimal acceptable levels of reliability (ICC = 0.79, lower 1-sided 95% CI 0.68). Conclusions: Both exercise diaries and self-rated scales demonstrated questionable validity in measuring home exercise adherence when compared with the objective accelerometer measured exercise completion. The self-rated adherence scale was also found to have limited test-retest reliability. Consequently neither of the adherence measures examined in this study can be confidently recommended for clinicians or researchers to use.