Abstract Background Accelerometer-measured physical activity is an increasingly used endpoint in heart failure (HF) trials. Purpose We aimed to investigate the determinants of accelerometer-measured physical activity and to establish the minimal clinically important difference (MCID). Methods Post-hoc analysis of the Empire HF trial, including stable outpatients with HF with reduced ejection fraction (HFrEF). Physical activity was quantified as average accelerometer counts per minute (CPM) with higher values representing higher activity. Associations between activity level and important clinical variables [age, sex, body mass index, N-terminal pro-brain natriuretic peptide, left ventricular ejection fraction (LVEF), atrial fibrillation, and anemia], as well as patient-reported health status (Kansas City Cardiomyopathy Questionnaire, KCCQ) were assessed. Results Complete data were available in 180 (95%) patients (86% male, mean age 65 years, mean LVEF 30%). Baseline median physical activity level was 1318 CPM (Q1-Q3 1111–1585). At baseline, age and anemia were independently associated with activity level (β-coefficients: -10 CPM per year age increase [95% CI -16 to -5.1], p = 0.00015, and -126 CPM for presence of anemia [95% CI -9.1 to -244], p = 0.035). Moreover, a significant independent association was observed between activity level and all KCCQ summary scores (β-coefficient point estimates of 3.7, 4.6, and 4.9 CPM, all p<0.02). For 12-week changes, only the KCCQ-clinical summary score (CSS) was associated with activity level, with a mean increase of 17.5 CPM [95% CI 1.5 to 34.0], p = 0.032, for every clinically relevant 5-point increase in KCCQ-CSS, corresponding to the MCID. Conclusion In patients with HFrEF, high age and anemia were independently associated with lower accelerometer-measured physical activity. Moreover, clinically relevant 5-point increases in Kansas City Cardiomyopathy Questionnaire clinical summary scores at 12 weeks were associated with an increase of 17.5 activity counts per minute, thus representing the minimal clinically important difference.Table 1.Baseline characteristicsFigure 1