Risk compensation is a theory which suggests that people typically adjust their behavior in response to perceived levels of risk, becoming more careful where they sense greater risk and less careful if they feel more protected. Although usually small in comparison to the fundamental benefits of safety interventions, it may result in a lower net benefit than expected or even higher risks.

By way of example, it has been observed that motorists drove closer to the vehicle in front when the vehicles were fitted with anti-lock brakes. There is also evidence that the risk compensation phenomenon could explain the failure of condom distribution programs to reverse HIV prevalence and that condoms may foster disinhibition, with people engaging in risky sex both with and without condoms.

By contrast, shared space is an urban street design method which consciously aims to increase the level of perceived risk and uncertainty, thereby slowing traffic and reducing the number and seriousness of injuries.

Risk compensation is related to the broader term behavioral adaptation which includes all behavior changes in response to safety measures, whether compensatory or not. However, since researchers are primarily interested in the compensatory or negative adaptive behavior the terms are sometimes used interchangeably. The more recent version emerged from road safety research after it was claimed that many interventions failed to achieve the expected level of benefits but has since been investigated in many other fields.

The reduction of predicted benefit from regulations that intend to increase safety is sometimes referred to as the Peltzman effect in recognition of Sam Peltzman, a professor of economics at the University of Chicago Booth School of Business, who published "The Effects of Automobile Safety Regulation" in the Journal of Political Economy in 1975 in which he controversially suggested that "offsets (due to risk compensation) are virtually complete, so that regulation has not decreased highway deaths". Peltzman claimed to originate this theory in the 1970s but it was used to oppose the requirement of safety equipment on trains in the Nineteenth Century.

A reanalysis of his original data found numerous errors and his model failed to predict fatality rates before regulation. According to Peltzman, regulation was at best useless, at worst counterproductive. Peltzman found that the level of risk compensation in response to highway safety regulations was complete in original study. But "Peltzman's theory does not predict the magnitude of risk compensatory behaviour." Substantial further empirical work has found that the effect exists in many contexts but generally offsets less than half of the direct effect. In the U.S., motor vehicle fatalities per capita declined by more than half from the beginning of regulation in the 1960s through 2012. Vehicle safety standards accounted for most of the reduction augmented by seat belt use laws, changes in the minimum drinking age, and reductions in teen driving.

The Peltzman effect can also result in a redistributing effect where the consequences of risky behaviour are increasingly felt by innocent parties (see moral hazard). By way of example, if a risk-tolerant driver responds to driver-safety interventions, such as compulsory seat belts, crumple zones, antilock brakes, etc. by driving faster with less attention, then this can result in increases in injuries and deaths to pedestrians.

Risk homeostasis is a controversial hypothesis, initially proposed in 1982 by Gerald J. S. Wilde, a professor at Queen's University in Canada, which suggests that people maximise their benefit by comparing the expected costs and benefits of safer and riskier behaviour and which introduced the idea of the target level of risk. He proposed four constituents to a person's calculations relating to risk: