This study describes a novel calibration method for six-degrees-of-freedom force/torque sensors (FTsensors) using a pre-calibrated force plate (FP) as a reference measuring device. In this calibration method, the FTsensor is rigidly connected to a FP and force/torque data are synchronously recorded while a dynamic functional loading procedure is applied by the researcher. Based on these data an accurate calibration matrix for the FTsensor can easily be obtained via least-squares optimization.Using this calibration method, this study further investigated what loading methods are appropriate for the calibration of FTsensors intended for ambulatory measurement of ground reaction forces (GRFs). Seven different loading methods were compared (e.g., walking, pushing while standing on the FTsensor). Calibration matrices were calculated based on the raw data from the seven loading methods individually and all loading methods combined. Performance of these calibration matrices was subsequently compared in an . in situ trial. During the . in situ trial, five common work tasks (e.g., walking, manual lifting, pushing) were performed by an experimenter, while standing on the FP wearing a "ForceShoe" with two calibrated FTsensors attached to its sole. Root-mean-square differences (RMSDs) between the FTsensor and FP outcomes were calculated over all tasks. Using the calibration matrices based on all loading methods combined resulted in small RMSDs (GRF: <8. N, center of pressure: <2. mm). Using the calibration matrices based on "pushing against manual resistance" resulted in similar RMSDs, proving it to be the best single loading method. © 2012 Elsevier Ltd.