Various methods have been published in the literature to estimate endocranial capacity. These are based on mathematical equations using measurements made directly on the skull or indirectly from X-rays, by filling the skull with various materials, by endocasts both physical and virtual (using 3D CT-scan reconstructions). Each method has its advantages, limits and drawbacks. A useful method is one that is simple, time-efficient, cheap, and reproducible. For this study we estimated endocranial capacity by filling the skull with calibrated glass beads. We used a thorough protocol of tamping to ensure results reproducibility. Intra- and inter-observer trials showed no significant differences. The method was applied to 130 recent French (West European) skulls (65 males and 65 females). Sexual dimorphism was studied using logistic regression functions, discriminant analysis, and neural network. The best results were obtained using the neural network, built with two hidden nodes in the hidden layer and one output node, and trained with a steepest descent algorithm. Our tests show that there is obvious sexual dimorphism (p < 10(-5)) of the weight of the skulls (623.44 ± 106.83 g in males; 548.07 ± 94.57 in females), the weight of the glass beads used to fill the skulls (2379.83 ± 228.92 g in males; 2095.95 ± 145.49 in females), and endocranial capacity (1676.47 cm(3) ± 161.26 in males; 1476.48 ± 102.49 in females), but overlapping between sexes is quite significant. The average ECC in male (1676.5 cm(3)) and in female (1476.5 cm(3)) subjects is higher than those previously published, which deserves some discussion.