It has been commonly argued that, in house mice, female post-partum fighting against a male intruder functions to protect the offspring from infanticide. The aim of this study was to test the hypothesis that maternal aggression is actually related to pup defence and, specifically, according to parental investment theory, that its intensity should increase with litter size. 60 nulliparous albino female mice were mated and randomly assigned to four experimental groups in which litters were culled at birth to 0, 4, 8, or 12 pups, respectively. On day 8 after delivery all females were tested for maternal aggression against a stranger adult male conspecific (5-min exposure). No aggression occurred in the group in which all pups had been removed. In the other groups, the proportion of females displaying overt aggression increased with litter size. Several scores of female agonistic behaviour (proportion of females displaying overt aggression, total attacking time, frequency of tail rattling) were significantly higher for the females rearing 8 and 12 pups than for the females rearing 4 pups. Aggressive behaviour of females rearing 12 pups was not significantly higher than that of females rearing 8 pups. No male committed infanticide. These results support the hypothesis that rodent maternal aggression is strictly related to offspring defence and are consistent with the theoretical prediction that, the costs of the defence being equal and the gain in fitness increasing with litter size, the intensity of maternal defence of the young should increase with their number.