Attentional bias plays an important role in the development and maintenance of alcohol addiction, and has often been measured with a visual probe task, where reaction times are compared for probes replacing either a substancerelated cue or a neutral cue. Systematic low-level differences between image classes are a potential cause of low internal reliability of the probe task (Ataya et al., 2012). Moreover, it is unclear whether automatic attentional capture by low-level properties such as size and colour in the non-substance related image could reduce attentional bias to the alcohol-related cue. Here, alcohol-related attentional bias was assessed in moderate social drinkers by measuring reaction times to targets that replaced either an alcohol-related or a non-alcohol related (i.e., neutral) picture. All alcohol-related images were greyscale, and the neutral stimulus could be either greyscale (‘control’), in colour (‘colour’), or greyscale and 25% larger in size (‘25% larger size’). We found attentional bias towards the alcohol-related stimuli in the control and 25% larger size conditions, but not in the colour condition. The magnitude of attentional bias was significantly reduced in the colour condition compared to the control and 25% larger size conditions. These findings indicate that salient low-level features in the non-substance related cue, in particular colour, can reduce the effect of alcohol-related content on the allocation of alcohol drinkers’ attention. Further, the results highlight the need for image pairs in visual probe tasks to be closely matched on basic perceptual dimensions.