Social groups of alpine marmots, Marmota marmota, often contain several adult males only one of which is a dominant territorial male. How reproductive competition may be expressed among adult group males during mating was investigated, in particular whether reproduction is suppressed in subordinate males. Among adult subordinate males at least 3 years old, potential offspring of the territorial male ('sons') had androgen levels as high as those of territorial males whereas non-related group males ('non-sons') had significantly lower vernal levels, similar to those of 2-year-old individuals. Plasma androgen and corticosteroid titres were negatively correlated in subordinates at least 2 years old but were positively correlated in territorial males. Corticosteroids correlated negatively with body mass in sons and were high regardless of body mass in non-sons. Androgens mirrored these relations. Injuries from intraspecific fighting were more frequent among non-sons than among sons. Androgens and corticosteroids of territorial males correlated positively with the number of non-sons in the group, whereas the number of sons in a group had no comparable or at best the opposite effect. These results indicate that territorial males attempt to inhibit reproduction in other group males but adjust their effort according to inhibition costs (presumably increasing with a subordinate male's body size) and benefits (which are lower in sons). Marmots with high corticosteroid levels in spring gained less mass during the following summer. This could impair winter survival or reproductive success in the next year. Body-mass loss during hibernation correlated positively with vernal androgen levels in adult males. Only males with sufficient summer mass gain may be capable of sustaining energetically costly interruptions of torpor at the end of winter, necessary for gonadal development. The interactions between group members around mating time may reflect general patterns throughout the year that lead to selective reproductive inhibition in the group via effects on mass gain, hibernation mass loss and gonadal development. (C) 1997 The Association for the Study of Animal Behaviour.