Purpose – This study aims to examine the construct validity and factor structure of the Rosenberg self-esteem scale (RSES) using a sample (n ¼ 312) of Polish prisoners incarcerated in Nowogard High Security Prison. Design/methodology/approach – The number of confirmatory factor analysis (CFA) models under investigation was limited to two by virtue of employing a much stricter and more rigorously sound methodological procedure in which item errors were prevented from correlating, as suggested by Brown. Findings – Confirmatory factor analyses indicated that the two-factor (positive and negative self-esteem) model provided a better fit for the RSES items than did the one-factor model. Originality/value – The results provide some initial support for the two-dimensional model that could possibly be measuring substantively separate factors within a prison sample, thus calling into question the one-factor solution of the RSES.