The genetic architectures of 12 behavioural variables measured in adult male mice placed in a novel environment were analysed in a replicated 4 X 4 diallel cross. The results were combined with those obtained in a classical cross involving two of the four strains. Based on the hypothesis of an evolutionary history of stabilising selection for mouse exploratory behaviour, we expected additive genetic effects and ambidirectional dominance. Such genetic architectures were actually found for those exploratory behaviours where epistatic effects were of minor importance. Similar findings emerged for some non-exploratory phenotypes. All behaviours analysed appeared to be polygenically controlled.