Introduction In anthropology, as in archaeology or forensic science, sexing is a very important part of any study, as many further analysis and interpretations will be based on it. Until now, there have been two principal methods to determine sex: morphognostic and metric, while analysis of DNA for sex determination is primarily used within a medico-legal context. Within the metric analyses repertoire there is the possibility of developing discriminant functions. This is done by deriving one or more equations from relevant measurements and obtaining a value which will act as cut-off point between males and females. The choice of measurements depends on two major factors: the degree of sexual dimorphism they present, and the level of preservation. Thus, some discriminant function analyses (DFA) rely on few measurements (e.g. Henke 1973, with a maximum of 5 measurements) or are based on specific anatomical structures which tend to be better preserved (e.g. Gapert et al. 2009, DFA for the occipital condyle). DFA for sex determination has been used on most parts of the skeleton with various degrees of success, varying between 60–90% (cranium: Gapert et al. 2009; Birkby 1966; Kajanoja 1966; Henke 1973; Ferembach et al. 1979; İşcan et al. 1995; Dayal et al. 2007; Franklin Bulletin der Schweizerischen Gesellschaft für Anthropologie 15(1–2): 57–63 (2009) Bulletin de la Société Suisse d’Anthropologie 15(1–2): 57–63 (2009) et al. 2005; mandibula: Calcagno 1981; Arnay de la Rosa et al. 2007; postcranium: Holland 1991; Marino 1995; Sacragi and Ikeda 1995; Dibbenardo and Taylor 1983), both for archaeological and forensic purposes. DFA can be used as a subsidiary method to morphognostic sex determination, but it has proved especially useful on isolated bones for which no standard visual method has been developed (Gapert et al. 2009; Marino 1995). Arnay de la Rosa (1997) even obtained more accurate sex determination results using metric analysis on a Canary Island population than with traditional visual methods. The advantage of metric analysis and the reason why it became increasingly popular is the „objectivity” of the method (Dayal 2007; Calcagno 1981), making obtained data easily comparable to other studies. One major drawback however, often pointed out, is the fact that the results are highly influenced by the size, robusticity and sexual dimorphism of the population for which the DFA had been developed (Marino 1995; Birkby 1966; Cowal and Pastor 2007; Walker 2008). Thus, Henke (1973) proposed to adapt DFA to other populations by shifting the sectioning point and Konigsberg et al. (2009) underlines the importance of knowing the ancestry before applying metrics on any individual or population. ISSN 14208-4835 Discriminant function analysis versus morphognostic sex determination of 637 crania from the Poschiavo ossuary GERALDINE D’EYRAMES 1 , FRANK SIEGMUND 2* & CHRISTINA P APAGEORGOPOULOU 1 1 Archaeological Service Graubünden; 2 Seminar für Ur- und Frühgeschichte, University of Basel; * Corresponding author Summary The paper presents a new discriminant function (DFA) for sex determination from cranial skeletal material specifically suitable for Swiss alpine populations. The formula was developed on a large cranial series (n = 637) originating from an ossuary in Poschiavo Switzerland (16th–19th century AD). The accuracy of the new DFA was compared to other published DFA (Giles and Elliot 1963; Henke 1973; Kajanoja 1966; Brùžek and Velemínský 2006) whereas a validation of all formulae was made on skeletal material mainly from Switzerland, southern Germany and Austria (12 series; ca. 1400 individuals; 9th mil. BC–19th century AD). The new DFA proved accurate with a high degree of concordance with morphognostic sex determination (79%) compared to the other DFA (69%–83%). Considering also that with the Poschiavo DFA more individuals could be classified based on the small number of necessary measurements, the present DFA proved especially valuable. The high accuracy levels and the fact that few measurements are needed make the proposed DFA suitable for Swiss alpine populations, applicable to a large amount of individuals and less time- consuming. Keywords: craniometrics, alpine population, ossuary, demography, Graubünden.