Sensitive to the many and varied aspects of risk-related behaviour, recent microeconomic developments in the study of uncertainty have increased the number of individual preference parameters ­ risk aversion, prudence, temperance, loss aversion, etc. ­ to cover it. Likewise, experimental and field studies have endeavoured to delimit this diversity by defining different types of risk ­ large and small risks, risks of gain and loss, with irreversible and reversible consequences ­ and to describe the behaviour relating to the area concerned (e.g. financial, professional and health) and context effects. Based on a specific questionnaire put to a sub-sample of the 1998 INSEE Patrimoine survey on personal wealth, which covers a wide range of areas of life, situations, contexts and types of risk, we paradoxically propose measuring this profusion of individual attitudes to risk using a single, purely ordinal indicator: we produce a score corresponding to all the responses given by each interviewee, as a summary measure supposed representative of the range of these preferences with regard to risk. A number of elements have been used to check the validity of this approach. If we go by the interviewees' own ranking of themselves on a scale of 0 (prudent) to 10 (adventurous), we find that they accept taking more risks when it comes to consumption and leisure and in their professional lives than with their health and family, but that the deviations remain small, with correlations of over 0.5 between the scales by area and the overall scale. More especially, the high degree of the score's internal consistency shows, ex post, the pertinence of an overall measure despite the diversity of attitudes to risk for any given individual.