Rationale: Following a mass casualty event, the primary psychological triage of survivors based on clinical symptomatology is difficult for volunteer rescuers who lack the necessary knowledge of mental health issues and are under significant time pressure. To address this concern, descriptive screening tools are needed. In this context, propositions for a triage tool have taken the form of means to describe traumatic exposure or the survivors' comorbidity factors. Objective: Alternatively, we propose a behaviour-based screening tool built on the Defence Cascade Model. This model assumes that survivors' behaviours involve a range of motor and physiological defensive reactions. After developing this new kind of screening tool, we tested its reliability and validity. Method: Forty professionals who usually intervene in early psychological intervention units randomly rated a series of survivors' profiles with either the behavioural tool or a tool based on classical categories of symptoms used as a control condition. The two screening tools were developed by a group of nine experts. Inter-rater reliability (unweighted kappa and Fleiss' Kappa indexes), criterion validity (Spearman's Rho and Kendall's Tau indexes), face validity (subjective evaluation) and construct validity (modal classes of items) were calculated and compared between the two tools. Results: The results show that behavioural screening is significantly more reliable, while showing a significant correlation with symptom-based screening in terms of prioritisation. It also appears to be simpler than symptom screening, with most references to similar clinical concepts.