Despite frequent reliance on surveys to document public attitudes towards conservation issues (such as invasive-species control), only rarely do researchers assess the validity of statements made by the public in response to such surveys. Therefore, how well responses match actual behaviour remains an open question. We conducted a survey asking drivers if they had seen and/or run over (intentionally or not) snakes, native frogs or invasive cane toads (Rhinella marina) on roads in the Northern Territory of Australia. To compare actual driver behaviour to the survey responses, we also carried out field experiments where we quantified the rates at which model snakes, frogs and toads (and controls) were run over on a rural highway. Our results show a discrepancy between survey responses and driver behaviour: for example, 25% of the people we surveyed indicated that they intentionally run over cane toads, yet field experiments showed that model toads were run over no more frequently than expected by chance, or than any other type of model.