The problem of game-vehicle accidents is discussed in terms of drivers' strategies for visual search in driving. To find possible measures for reducing the number of wildlife accidents, four studies were undertaken. Initially, two exploratory investigations were made: a survey of drivers' expectancies concerning moose in traffic and a study involving self reports of accidents and near accidents with moose. The results from these investigations give no evidence that drivers' experience, knowledge, orattitudes concerning moose are related to wildlife accidents.Instead they suggested that the visualsearch patterns of drivers might explain some of the effects obtained. In a series of field experiments, drivers' ability to detect moose dummies was explored, and in a final study, the effectiveness of the game crossing sign was investigated experimentally. The rsults were interpreted as evidence that in rural driving, drivers normally scan the view ahead in a systematic and almost automatic way which is not effective for the task of detecting moose. Drivers can easily change their automatic scanning into controlled search for animals but this search is demanding and can probably not be sustained for any length of time without feedback.