Introduction: Acceleration sensors are increasingly used for the assessment of physical activity during free-living. Acceleration sensor calibration is a potential source of error. This study aims to describe and evaluate an auto-calibration method to minimize calibration error using segments within the free-living records (no extra experiments needed). Methods: The auto-calibration method entailed the extraction of non-movement periods in the data, for which the measured vector magnitude should ideally be the gravitational acceleration (1g) gravity; this property was used to derive calibration correction factors using an iterative closest-point fitting process. The reduction in calibration error was evaluated in data from four cohorts: UK (N=921), Kuwait (N=120), Cameroon (N=311), Brazil (N=200). Results: Our method significantly reduced calibration error in all cohorts (p<.01), ranging from 16.6 to 3.0mg in the Kuwaiti cohort to 76.7 to 8.0mg error in the Brazil cohort. Utilising temperature sensor data resulted in a small non-significant additional improvement (p>.05). Temperature correction coefficients were highest for the z-axis, e.g. 19.6mg offset per 5⁰C. Further, application of the auto-calibration method had a significant impact on typical metrics used for describing human physical activity, e.g. in Brazil average wrist acceleration was 51% lower than uncalibrated values (p<.01). Discussion/Conclusion: The auto-calibration method as presented helps reduce the calibration error in wearable acceleration sensor data and improves comparability of physical activity measures across study locations. Temperature utilisation seems essential when temperature deviates substantially from the average temperature in the record, but not for multiday summary measures.