Background and objectives. Safety education often targets parental risk per
ception. Predictors of risk perception, however, are not well known, thus l
imiting the feasibility of effective safety education. Accordingly, in this
study, a range of predictors of maternal risk perception were examined. Me
thods: A random sample of 870 mothers in northern Sweden was included in th
e study. Three different questionnaires, with scenarios of a burn injury, a
bicycle injury in the home environment, and a bicycle injury in traffic, w
ere completed by the subjects. Multiple Linear regression models tested the
possible influence of causal attributions, normative beliefs, and sociodem
ographic and behaviour-related variables an mothers' risk perception. Resul
ts: Only 14-23% of the variance in mothers' risk perception could be explai
ned by the multivariate models. Causal attribution to the child was found t
o be the most important predictor of maternal risk perception. Conclusion.
Present theoretical models give few clues about how to design educational m
odels that might influence risk perception. To make safety education more e
ffective, other modifiable factors that influence parental safety behaviour
, such as subjective norms and self-efficacy, might be better targets.