Abstract: Risk-taking behaviours among adolescents have become a clinical reality and are related to a significant mortality and morbidity. Many of these behaviours are ordalique ones and consist in braving death with repetitive and deliberate risk-taking. Adolescents’ll leave their life to chance, fate, supernatural or divine powers in order to know if they must die or not. If subjects survive to fatal situations, they have a deep conviction that they’re protected and they feel therefore invincible. The ordalique behaviour is not equivalent to suicide. It’s a pathological way to restore a serious lack of narcissism and identity deficiencies. The ordalique functioning is a main characteristic of addictions, especially drug addiction. This article presents the construction of the ordalique functioning questionnaire (QFO) and initial validity studies. The QFO is composed of 71 items and four dimensions are evaluated : risk-taking, transgression, positive representation of risk-taking and believes. First dimension refers to the propensity of having dangerous activities (extreme sports, drugs use, risky sexuality, dangerous driving…). Second dimension refers to the subject’s relation to law, rules, prohibition and authorities and his tendency to have transgression behaviours. Third dimension refers to heroic representation of subjects who have risk-taking behaviours. This dimension mainly applies to narcissistic function of risk-taking. Fourth dimension refers to various believes (luck, fate, chance, god…) and the implication of these believes in risk-taking behaviours. In a first study, QFO was administrated to a population of 826 students from several universities of Toulouse. Four hundred and two of them were men and 424 were women. The age ranged from 18 to 37 with a mean of 21,13. The component analysis with varimax rotation allowed to find the four factors of construction. Alpha’s Cronbach are >0.60 for the four factors (from 0.73 to 0.79) and therefore underline the internal consistency of the questionnaire. There’s a good inter-correlation between the factors (from 0.44 to 0.60) except for believes which is not significantly correlated with the other three factors. Each result shows that believes could be an independent dimension. A test-retest study after an interval of 15 days among 80 students indicates excellent temporal reliability (from 0.93 to 1). In a second study, QFO was administrated to 150 subjects (60 drugs addicts, 30 restrictive anorexics, 60 controls). The 60 drugs addicts (30 men and 30 women with a mean age of 27, 28) were in aftercare centers. They have a very strong dependence on drugs and alcohol according to the Addiction Severity Index (ASI, McLellan) and all of them are polyaddicted. The 30 women anorexics (mean age of 19,50) were restrictive type according to the anorexia and bulimia sections of the MINI (Mini International Neuropsychiatric Interview, Sheehan and al.). Twenty anorexics were hospitalized during the research, four anorexics were met in childhood guidance structures and five anorexics were having psychotherapy treatment. The mean weight is 37,12 kg and the mean body mass index (BMI) is 14,67. The 60 controls are composed of 30 men and 30 women with a mean age of 23,55. Thirty eight of them are students and 22 are working people. The controls don’t have any psychiatric troubles listed in DSM-IV (according to the 17 diagnostic sections of the MINI), don’t smoke more than 10 cigarettes per day, don’t suffer from eating disorders and don’t have present problems with drugs or alcohol. Anova analysis and Mann Whitney tests were used to compare means between the three groups. Results of drug addicts are significantly higher than controls on risk-taking (U=0 ; P <0.0000), transgression (U=0 ; P <0.0000), positive representation of risk-taking (U=240 ; P <0.0000) and believes (U=15,5 ; P <.0000). Drug addicts also present results significantly higher than anorexics on risk-taking (U=1 ; P <0.0000), transgression (U=0 ; P <0.0000), positive representation of risk-taking (U=0 ; P <0.0000) and believes (U=11 ; P <0.0000). Anorexics have results significantly lower than controls on risk-taking (U=216,5 ; P <0.0000), transgression (U=288 ; P <0.0000), positive representation of risk-taking (U=217 ; P <0.0000) and believes (U=118 ; P <0.0000). These results underline in drug addicts an important psychic ordalique functioning that we can illustrate for example by the “russian roulette” with overdose, a risky sexuality and the share of syringes despite the threat of VIH. The very important differences between drugs addicts and anorexics indicate that anorexics would have a subjective relation to risk-taking different from drugs addicts. Indeed, unlike drug addicts, anorexics are not aware of the immediate and fatal risk of their behaviour. They don’t have a heroic representation of their risk-taking. They have not other risky activities apart from addiction. Anorexics don’t transgress laws and don’t have delinquent behaviours, they are rather conformist. They have few believes in chance, fate, destiny or god. Their results lower than controls on all dimensions of QFO showed that anorexics avoid sensations seeking and dangerous situations. The seeking of an ideal body is more important than braving death which is not even perceived. So, the subjective relation to risk-taking in anorexics wouldn’t be ordalique but “sacrificial”, that is to say focused on pathological self-control. These differences of functioning (ordalique versus sacrificial) between drug addicts and anorexics must be taken into account for therapeutic care. In this way, our results support the thesis of the existence of two types of addictions : those focused on sensations seeking, impulsivity, dangerous activities and those focused on sensations avoiding, pathological self-control, and no attraction for risk-taking. Moreover, this research shows that addictions could be classified according to a continuum from the more ordalique dependences (drug addiction) to the more accepted and passive ones (tobacco dependence) with at the other extremity the more sacrificial dependences (anorexia). In this second study, the Sensation Seeking Scale (SSS, Zuckerman) was also administrated. Correlations between SSS and QFO (from .70 to .87) indicated a very good concurrent validity. Moreover, there’s a very good inter-correlation between the four dimensions of QFO (from .76 to .90). These results, more satisfactory than in first study, can be due to the population which isn’t a run of the mill population but a pathological one with drugs addicts who have an important ordalique functioning. And notably, believes dimension, central in ordalique behaviours, is here strongly correlated with the three others dimensions of QFO. Therefore our thesis indicated above of an independent dimension can be excluded. In conclusion, QFO appears to be a promising tool for research in psychopathology. [Copyright &y& Elsevier]