Zebrafish are a widely utilised animal model in developmental genetics, and owing to recent advances in our understanding of zebrafish behaviour, their utility as a comparative model in behavioural neuroscience is beginning to be realised. One widely reported behavioural measure is the novel tank-diving assay, which has been often cited as a test of anxiety and stress reactivity. Despite its wide utilisation, and various validations against anxiolytic drugs, reporting of pre-test housing has been sparse in the literature. As zebrafish are a shoaling species, we predicted that housing environment would affect their stress reactivity and, as such, their response in the tank-diving procedure. In our first experiment, we tested various aspects of housing (large groups, large groups with no contact, paired, visual contact only, olfactory contact only) and found that the tank diving response was mediated by visual contact with conspecifics. We also tested the basal cortisol levels of group and individually housed fish, and found that individually housed individuals have lower basal cortisol levels. In our second experiment we found ethanol appeared to have an anxiolytic effect with individually housed fish but not those that were group housed. In our final experiment, we examined the effects of changing the fishes' water prior to tank diving as an additional acclimation procedure. We found that this had no effect on individually housed fish, but appeared to affect the typical tank diving responses of the group housed individuals. In conclusion, we demonstrate that housing represents an important factor in obtaining reliable data from this methodology, and should be considered by researchers interested in comparative models of anxiety in zebrafish in order to refine their approach and to increase the power in their experiments.