III response to disagreements about the extent to which smokers recognize t
he full risk: of smoking-induced illness, an attempt was made to review all
articles that have investigated smokers' risk perceptions. These diverse s
tudies are grouped here into four categories, depending oil the type of ris
k judgment solicited by researchers. This grouping shows that the apparent
underestimation or overestimation of risk depends on how risk perceptions a
re assessed. No single conclusion about the accuracy of smokers' numerical
risk estimates is possible since the accuracy depends entirely oil the heal
th outcome rated (e.g. lung cancer versus all deaths duc to smoking). With
other types of risk questions, smokers consistently acknowledge that smokin
g increases health risks, but they judge the size of these increases to be
smaller and less well-established than do non-smokers. Finally smokers mini
mize the personal relevance of the risks: they do not believe that they are
as much at risk as other smokers of becoming addicted or suffering health
effects. The accumulated data indicate that smokers continue to minimize th
eir personal health risks.