The theory of rational decision-making models risk in terms of mathematical expectation, taking into account the probability and magnitude of the outcome. However, risk assessment decisions cannot be predicted when, as in the case of AIDS, decisions often are based on inadequate or biased information. To explain deviances rational risk assessment decision, researchers have argued that when faced with complex risk choices, individuals may use a variety of cognitive heuristics to simplify the choice problem. For example, if personal experience as a source of likelihood information is missing, individuals will rely on press and media coverage, which is often biased. Representativeness heuristics reduce probability judgements to judgements of similarity. Finally, individuals often consider risk attributes other than probability and outcome; of particular interest in the case of AIDS risk is the concept of voluntary versus involuntary risks. Self-assessed knowledge about AIDS is greater among high-risk groups, specifically homosexuals. Although advertising and health promotion does improve knowledge about AIDS, it can only be considered an intermediate measure of public health effect. Policies to provide information on AIDS must be evaluated in terms of their ability to change individuals' perceptions of their personal risk and the value they place on precautionary measures. 2 tables, 23 references. (Author abstract modified)