The study examined the psychometric properties of scores on the English version of the short Physical Self-Inventory (PSI-S) among 1368 English-Speaking and 224 French-Speaking adolescents. Participants completed the original PSI-S and a revised version including positively-worded reformulations of the original negatively-worded items. The results supported the a priori factor structure, composite reliability, and test-retest reliability (7-8 months) of scores on the revised PSI-S, and its equivalence with the French version. Compared to confirmatory factor analyses, exploratory structural equation modeling resulted in a superior solution, and more cleanly differentiated factors. PSI-S responses were longitudinally invariant across 7-8 months, and presented no evidence of differential item functioning as a function of age, gender, body mass index (BMI) and physical activity. The results showed latent mean differences as a function of gender, BMI (linear and curvilinear effects), and physical activity, supporting the criterion-related validity of the PSI-S factors. (C) 2016 Elsevier Ltd. All rights reserved.