ABSTRACT: Many studies have established that threatening stimuli (such as spiders, snakes and weapons) tend to be located more quickly in visual search tasks than superficially similar, but not threatening, stimuli (such as beetles, lizards, tools and gadgets). Response times in visual search tasks can therefore be used as an index of threat evaluation and attentional priority. However, many other aspects of visual search stimuli influence response times. Low level characteristics, such as luminance, contrast and colours within a target, and differences between targets and distracters are important. In addition, attentional priority may be given to some types of stimuli over others, for reasons other than the threat level they imply. In this presentation I will describe a new behavioural measure of responding for visual search tasks that attempts to separate the effects of threat evaluation from low-level stimulus properties on participants' response times in these tasks. This ‘caution score’ increases as threat level of targets increases (from beetles, to non-lethal spiders, to lethal spiders, to lethal spiders depicted in peri-personal space), does not seem to be affected by changes in the threat level of distracter stimuli, increases when weapons are depicted wielded compared to when they are not, and has also revealed sex difference, with males responding to weapons, but no other target stimuli, more cautiously than females. I will also discuss the limitations of this caution score and some methodological changes that might increase its sensitivity.