The concept of risk, in relation to human health, is a topic of primary interest for occupational health professionals. A new legislation recently established in Italy (626/94) according to European Community directives in the field of Preventive Medicine, called attention to this topic, and in particular to risk assessment and evaluation. Motivated by this context and by the impression that the concept of risk is frequently misunderstood, the present paper has two aims: the identification of the different meanings of the term "risk" in the new Italian legislation and the critical discussion of some commonly used definitions; and the proposal of a general definition, with the specification of a mathematical expression for quantitative risk estimation. The term risk (and risk estimation, assessment, or evaluation) has mainly referred to three different contexts: hazard identification, exposure assessment, and adverse health effects occurrence. Unfortunately, there are contexts in the legislation in which it is difficult to identify the true meaning of the term. This might cause equivocal interpretations and erroneous applications of the law because hazard evaluation, exposure assessment, and adverse health effects identification are completely different topics that require integrated but distinct approaches to risk management. As far as a quantitative definition of risk is of concern, we suggest an algorithm which connects the three basic risk elements (hazard, exposure, adverse health effects) by means of their probabilities of occurrence: the probability of being exposed (to a definite dose) given that a specific hazard is present (Pr(e[symbol: see text]p)), and the probability of occurrence of an adverse health effect as a consequence of that exposure (Pr(d[symbol: see text]e)). Using these quantitative components, risk can be defined as a sequence of measurable events that starts with hazard identification and terminates with disease occurrence; therefore, the following formal definition of risk is proposed: the probability of occurrence, in a given period of time, of an adverse health effect as a consequence of the existence of an hazard. In formula: R(d[symbol: see text]p) = Pr(e[symbol: see text]p) x Pr(d[symbol: see text]e). While Pr(e[symbol: see text]p) (exposure given hazard) must be evaluated in the situation under study, two alternatives exist for the estimation of the occurrence of adverse health effects (Pr(d[symbol: see text]e)): a "direct" estimation of the damage (Pr(d[symbol: see text]e) through formal epidemiologic studies conducted in the situation under observation; and an "indirect" estimation of Pr(d[symbol: see text]e) using information taken from the scientific literature (epidemiologic evaluations, dose-response relationships, extrapolations, ...). Both conditions are presented along with their respective advantages, disadvantages, and uncertainties. The usefulness of the proposed algorithm is discussed with respect to commonly used applications of risk assessment in occupational medicine; the relevance of time for risk estimation (both in the term of duration of observation, duration of exposure, and latency of effect) is briefly explained; and how the proposed algorithm takes into account (in terms of prevention and public health) both the etiologic relevance of the exposure and the consequences of exposure removal is highlighted. As a last comment, it is suggested that the diffuse application of good work practices (technical, behavioral, organizational, ...), or the exhaustive use of check lists, can be relevant in terms of improvement of prevention efficacy, but does not represent any quantitative procedure of risk assessment which, in any circumstance, must be considered the elective approach to adverse health effect prevention.