To determine the ActiGraph GT3X+ cut-points with the highest accuracy for estimating time spent in sedentary behaviour in older adults in free-living environments. ActivPAL 3 ™ was used as the reference standard.Cross-sectional study.37 participants (13 males and 24 females, 73.5±7.3 years old) wore an ActiGraph GT3X+ and an ActivPAL 3 ™ for 7 consecutive days. For ActivPAL 3 ™, variables were created based on posture. For ActiGraph GT3X+, sedentary behaviour was defined as (1) vector magnitude and (2) vertical axis counts for 1-s, 15-s and 1-min epochs, with cut-points for 1-s epochs of <1 to <10counts, for 15-s epochs of <1 to <100counts and for 1-min epochs of <1 to <400counts. For each of the ActiGraph GT3X+ cut-points, area under the receiver operating characteristic curve (area under the curve), sensitivity, specificity, and percentage correctly classified were calculated. Bias and 95% limits of agreement were calculated using the Bland-Altman method.The highest areas under the curve were obtained for the vector magnitude cut-points: <1count/s, <70counts/15-s, and <200counts/min; and for the vertical axis cut-points: <1count/s, <10counts/15-s and <25counts/min. Mean biases ranged from −4.29 to 124.28min/day. The 95% limits of agreement for these cut-points were ±2h suggesting great inter-individual variation.The results suggest that cut-points are dependent on unit of analyses (i.e. epoch length and axes); cut-points for a given epoch length and axis cannot simply be extrapolated to other epoch lengths. Limitations regarding inter-individual variability and misclassification of standing activity as sitting/lying must be considered.