Objectives Sex‐specific differences in the shape of the iliac crest have been reported based on quantification of simplified curvature using two‐dimensional variables or qualitative identification of faint/marked S‐shaped curvature. An objective and quantitative approach for the evaluation of iliac crest curvature in juveniles was developed. Using three‐dimensional data, ontogenetic variation in iliac curvature was investigated in relation to size and chronological age and its accuracy for sex estimation among juveniles was evaluated. Materials and Methods Three‐dimensional geometric morphometric landmarks were collected from virtual models of 53 surface‐scanned ilia and used to: (1) quantify differences between male and female shape, and (2) to calculate a measure of sinuosity by extracting linear measurements relative to a plane constructed in virtual space. Results Males were found to exhibit on average more pronounced curvature of the iliac crest than females in equivalent age groups, and the difference between sexes was more marked on the posterior section of the bone than on the anterior section. Classification accuracy was higher for measurements of sinuosity relative to a three‐dimensional plane (62–87%) than for landmarks describing crest shape (64–74%). Conclusions An increased likelihood of correct male identification (reduced rate of false positive identification) was observed in older children (>5 years). This reflects divergence of male and female iliac crest shape with size (or age), and the development of a unique, more curved iliac crest shape in males. The results suggest more conspicuous male trait expression than female trait expression.