The aim of this study was to test the reliability of Osborne et al. (2004) [8] age estimation based on the iliac auricular surface of the ilium. We selected 172 skeletons with documented sex and age-at-death and established six uniformly distributed age intervals for analysis. ANOVA was employed to assess the association of the documented age with sex and the auricular surface classification. We employed Bonferroni post-hoc tests to find any statistical differences across documented ages within each phase of Osborne's criteria. While the ANOVA showed a significant association between the documented age and the auricular surface classification, post-hoc tests found that phases 2 and 3 were the only consecutive phases with significant differences. We argue that a lack of statistical significance between consecutive phases undermines the reliability of this method for forensic purposes especially in middle adults. It may be possible that the collapse of intermediate phases would increase the accuracy of this method. (C) 2017 Elsevier B. V. All rights reserved.