A review of the literature on chlordiazepoxide indicates that the traditional view of benzodiazepines as antiaggressive drugs represents an inaccurate generalization. In fact, highly variable findings have been reported with the most significant factor in the variability being the type of aggression studied. Furthermore, considerable controversy surrounds the nature of chlordiazepoxide's influence on social conflict in rodents, with opinion divided regarding the selectivity of its inhibitory effect on offensive responding. It is argued that inappropriate behavioural methodology may have substantially contributed to this controversy. A pharmacoethological approach to the analysis of drug effects on social behaviour is described and exemplified by new data on chlordiazepoxide and midazolam. Both agents inhibit offense at doses which do not result in general response inhibition, yet their overall behavioural profiles are somewhat different. Evidence is also presented indicating low dose behavioural activity of the benzodiazepine antagonist Ro15-1788 in two test situations and suggesting possible differences in the effects of Ro15-1788 and CGS8216 on novelty-related responding. Several avenues of research are discussed which may yield insights into the manner whereby benzodiazepines influence social patterns and the significance of benzodiazepine-GABA interactions in such processes.