This paper is concerned with evaluation of the measurement uncertainty of true stress-true strain curves of auto body steel sheets at various strain rates considering correlation of input quantities. The measurand is defined as the true stress at strain rates ranging from 1 to 100 s(-1). The true stress has a functional relation with the tensile load, the initial width and the initial thickness, the initial and deformed length of a specimen. Since the initial and deformed lengths of a specimen are measured by the exactly same procedure, the two quantities are correlated and considered in the uncertainty evaluation model. An analytic model to evaluate the measurement uncertainty is established properly by considering the correlation with the guidelines suggested by the GUM, ISO/IEC Guide 98-3. The standard uncertainties of the input quantities and influence factors are evaluated for dynamic tensile properties of DDQ, TR1P980, and TRIP1180 steel sheets as well as the covariance associated with the correlated input quantities. The standard uncertainty of the true stress is also evaluated to consider the change of the strain rate during the test. The combined standard uncertainty is then evaluated including standard uncertainties of the input quantities and influence factors. The expanded uncertainty is obtained by choosing an appropriate coverage factor. The absolute amount of the measurement uncertainty decreases with consideration of the correlation between the initial and deformed length since their uncertainty contributions are almost canceled during the calculation of combined uncertainty. Finally, the reliability of dynamic tensile properties of auto-body steel sheets is presented with a statement of the expanded uncertainty. (C) 2017 Elsevier Ltd. All rights reserved.