This study examined the accuracy and reproducibility of a body-fixed sensor-based assessment for quantifying frontal plane angular movements of the (upper) thorax and pelvis of patients with hip osteoarthritis at different walking speeds. To evaluate accuracy, the angular movements of sensors attached to the thorax and pelvis of three patients were compared with results based on an optical motion analysis system. Accuracy was high, with small and consistent mean differences (<1.0 degree) and corresponding standard deviations (<1.3 degrees) between optical motion analysis system and body-fixed sensor data. To evaluate reproducibility, angular trunk movements were assessed twice in 15 patients. Reproducibility was high (intraclass correlation coefficients ranged from 0.86 to 0.97), and the values of the mean differences between the test and retest were small, with the 95% confidence interval containing zero. This body-fixed sensor-based assessment is an accurate and reproducible method for quantifying frontal plane compensatory trunk movements during gait of patients with hip osteoarthritis at different walking speeds.