Risks - and particularly health risks - are predictions about outcomes of human actions. However, actions always have multiple outcomes, and simplistic assertions that a behaviour is 'risky' discounts the multiplicity of ways in which outcomes are differentially valorised. The paper looks at three models of the relationship between risks and hazards, and concludes that hazards are constructs generated from narrow perspectives on the consequences of action. Acknowledging this, the paper goes on to look at people's behaviour as'experiments in living'. Asking'what can a body do?' is a way to assess the possibilities for such experiments. Looking at the use of the recreational drug Ecstasy, it is argued that drug users apply a knowledge ability of what a body can do which derives from positive assessments of Ecstasy's effects rather than medical knowledge abilities. What appears to be resistance to research evidence about behaviour is actually pressure against normative limits of what a body can do.