Introduction: Assessment of physical activity is an important outcome requiring valid and reliable measures for physical activity (PA) research in patients with heart failure (HF). Purpose: The purpose of this study was to assess reliability of two accelerometer measures of PA using Intraclass Correlation Coefficients (ICCs) and to compare PA levels from self-report and physiological assessment with accelerometers in patients with HF. Daily physical activity levels were compared for self-reported PA from the Modified 7 Day Physical Activity Recall (7 Day PAR) questionnaire and two accelerometry measures (RT3 and Actiheart). Methods: A cross-sectional descriptive design was used in a convenience sample of 31 elderly HF patients. Subjects had systolic HF with a mean ejection fraction of 29.1 6 7.3% and a mean body mass index of 29.3 6 5.6. Twenty-one males and ten females with a mean age of 72.5 6 7.1 years wore the accelerometers for 7 days. The 7 Day PAR was completed by interview at the end of the week. Results: The RT3 and Actiheart had ICC reliability estimates of 0.92 and 0.83 respectively. Subjects were sedentary a mean of 96.0% of the time (23 of 24 hours per day). Sedentary activity ranged from 69.0% to 99.9% of the 24 hour day in this sample. Mean kilocalories per day spent in activity was not significantly different [t (df 27) 5 -0.99, pO0.05] between the two accelerometers. The Actiheart showed a mean of 355.9 6 205.5 activity kilocalories per day as compared to the RT3 with 311.8 6 215.3 activity kilocalories per day. Self-reported mean kilocalories per day from the 7 Day PAR was significantly (p!.001) higher (1221.5 6 814.2 kcal) as compared to each of the accelerometers. Conclusions: Results from this study show that both accelerometers had acceptable reliability estimates with 7 days of measurement. Elderly HF patients are sedentary a majority of each 24 hour day and they consistently over-estimate physical activity with self-report as compared to objective measurement with accelerometry. Implications: Research to assess PA outcomes should not rely on self-report measures but must include a physiological measurement. Seven days of measurement with accelerometry provided acceptable reliability estimates in this elderly HF population.