Various scholars have pioneered sex assessment from different parts of the skeleton based on metrical or qualitative morphological characteristics. Lately though, a new technique which combines both traditional methods is becoming popular in forensic anthropology. Geometric-morphometrics is a field of multivariate statistical biometric analysis which allows the quantification of the shape and size components of morphological variation. The purpose of this investigation is to assess sex from the humerus in a contemporary Greek population using geometric-morphometric analysis of shape data derived from digital radiographs. The study population consists of 97 well-preserved adult humeri from two cemeteries of Heraklion, Crete. The left humeri are radiographed using digital radiograph machine (TCA 4R PLUS). Assuming fragmentary patterns, the proximal and distal ends are studied separately. Five landmarks are selected on the radiograph of the proximal epiphysis and seven landmarks on the distal. Generalised Procrustes analysis (GPA) and thin-plate splines are used to obtain the shape and size variables for statistical analysis. Then three discriminant function analyses were carried out: one uses the PC scores from Procrustes shape space, the second the centroid size alone and the third the PC scores of GPA residuals plus lnCS for analysis in Procrustes form space. Results indicate the existence of shape differences between the sexes. In females the greater tubercle is smoother, with its superior border less pronounced. Additionally, females have a relatively squared distal epiphysis, while males exhibit a more rectangular shape. Shape differences between genders, for the cross-validated data, give slightly better classification results in the proximal humerus (73%) compared with the distal humerus (71%). Size alone performed better (86.5% for proximal and 85.6% for distal humerus). As anticipated, the classification accuracy improves (89.6% for proximal and 89.7% for distal epiphysis) when both size and shape are combined. From the forensic standpoint, the usefulness of this study rests on the identification of sex based on shape differences observed on radiographs of fragmentary humeri that could not be assessed with traditional methods. The analysis of humeral radiographs by geometric-morphometric techniques offers an alternative way to identify the sex of unknown skeletal remains. Whether this is statistically better than simply using traditional osteometric methods is a question that needs to be tested in a meta-statistical approach.