We present a marginal value model explaining intraspecific and interspecific variation of mating systems in nonterritorial ungulates. The model takes into account the simultaneous effects of spatial and temporal distribution of females, female group size, male-male competition, female choice, and the operational sex ratio (i.e., the proportion of estrous females). The model predicts that higher numbers of females per group increases the average exploitation time of such groups by males. An increase in female group density, operational sex ratio, and age-specific fighting success of males are predicted to reduce the average exploitation time of female groups, leading to roving of males (i.e., moving between female groups). In contrast, an increase in die female rejection rate of males and in the time spent by males on nonmating activities (i. e., foraging, lying down, ruminating, migrating) are predicted to increase the average exploitation time of female groups and to favor staying behavior of males (i.e., defending a female group over a longer period of time). Consequently, die model predicts that young males will tend to be “stayers,” whereas middle-aged and old males are expected to be “rovers.” Model predictions were tested widi field data collected on muskoxen Ovibos moschatus in a natural population in west Greenland. Observed correlations between bull exploitation times of cow groups and the six above-mentioned social and environmental parameters were all in die predicted direction and statistically significant in five of die six cases. Overall, 69% of die observed variation in exploitation time of cow groups by males was explained by die model predictions. Stepwise regression suggested that, of die six parameters, variation in sex ratio (R2 = .56) and time spent on nonmating activities (R2 = .35) had the largest effects on male exploitation time. Also, die observed age-specific variation in bull exploitation time of cow groups was as predicted.