Behavioural avoidance tests (BATs) are a cornerstone of objective assessment of phobias. However, live BATs have several disadvantages. They are practically difficult and time-consuming to set up and are not standardised. This study examined two computer-delivered BATs (using slide and video presentations of phobic stimuli respectively): first, in respect to their ability to discriminate fearfuls from nonfearfuls, and second, in terms of convergent validity with a live BAT and the Spider Phobia Questionnaire (SPQ). Sixty-four low (n = 32) and high (n = 32) spider-fearful undergraduate participants were administered the three BATs in counterbalanced order. Results showed that subjective anxiety on all BATs was highly discriminative of low and high spider-fearfuls. The number of steps completed did not discriminate between phobics and nonphobics on the computer BATs. However, there was good convergent validity between the live BAT, the SPQ and both computer-delivered BATs on subjective anxiety. Overall, the live BAT gives a clearer indication of avoidance behaviour while the video BAT assesses subjective anxiety across a wider range of steps. The development of computer-delivered BATs that reliably measure avoidance is necessary before contemplating them as an alternative to a live BAT.