Cortical bone surface area estimation is generally performed by analysis of medical images but its accuracy has not yet been verified by direct bone measurements. This study was performed to evaluate the reproducibility of direct ex vivo estimations of cross sectional tibial cortical bone surface area. Ten tibiae were evaluated by two observers using direct 3D-digitization at 38 and 66% of total tibial length using a 'critical' and a 'general' criterion for differentiation between cortical and cancellous bone. The results of the analysis using the more severe 'critical' differentiation criterion yielded high intra and interobserver agreement. Mean intraobserver reliability was demonstrated by R = 0.98 (Spearman rank correlation between 0.98 and 0.97; P < 0.01) and inter observer reliability by R = 0.91 (Spearman rank correlation between 0.95 and 0.88; P < 0.01). When using strict criteria to differentiate between cortical and cancellous bone, intra and interobserver reliability of the presented method of direct cortical bone surface area estimation is very high. The results indicate that cross-sectional cortical bone surface estimation by direct 3D-digitization can serve as a gold standard for validation of other estimations methods.