Many animal species exhibit year-round aggression, a behaviour that allows individuals to compete for limited resources in their environment (eg, food and mates). Interestingly, this high degree of territoriality persists during the non-breeding season, despite low levels of circulating gonadal steroids (ie, testosterone [T] and oestradiol [E2 ]). Our previous work suggests that the pineal hormone melatonin mediates a 'seasonal switch' from gonadal to adrenal regulation of aggression in Siberian hamsters (Phodopus sungorus); solitary, seasonally breeding mammals that display increased aggression during the short, 'winter-like' days (SDs) of the non-breeding season. To test the hypothesis that melatonin elevates non-breeding aggression by increasing circulating and neural steroid metabolism, we housed female hamsters in long days (LDs) or SDs, administered them timed or mis-timed melatonin injections (mimic or do not mimic a SD-like signal, respectively), and measured aggression, circulating hormone profiles and aromatase (ARO) immunoreactivity in brain regions associated with aggressive or reproductive behaviours (paraventricular hypothalamic nucleus [PVN], periaqueductal gray [PAG] and ventral tegmental area [VTA]). Females that were responsive to SD photoperiods (SD-R) and LD females given timed melatonin injections (Mel-T) exhibited gonadal regression and reduced circulating E2 , but increased aggression and circulating dehydroepiandrosterone (DHEA). Furthermore, aggressive challenges differentially altered circulating hormone profiles across seasonal phenotypes; reproductively inactive females (ie, SD-R and Mel-T females) reduced circulating DHEA and T, but increased E2 after an aggressive interaction, whereas reproductively active females (ie, LD females, SD non-responder females and LD females given mis-timed melatonin injections) solely increased circulating E2 . Although no differences in neural ARO abundance were observed, LD and SD-R females showed distinct associations between ARO cell density and aggressive behaviour in the PVN, PAG and VTA. Taken together, these results suggest that melatonin increases non-breeding aggression by elevating circulating steroid metabolism after an aggressive encounter and by regulating behaviourally relevant neural circuits in a region-specific manner.