I assumed that the reproductive strategies of the two Accipiter species investigated in this study, the northern Goshawk (A. gentilis) and the Cooper's hawk (A. cooperii), would be similar to the strategies of other monogamous raptors with sex role partitioning. Any differences in the species' reproductive strategies, particularly hunting strategies, were expected to be a function of their size differences. Contrary to predictions, results indicated that, in north-central New Mexico during brood rearing, both species had similar hunting strategies and utilized similar prey species. These results also indicated that the reproductive strategies of the Cooper's hawk were not adequately described by simple models of sex role partitioning. In this population, over 50% of the female Cooper's hawks deserted during the fledgling-dependency period and made no attempt to renest. In all cases of desertion, the male provided uniparental care until the young were independent. Based on time-energy budget estimates, the deserting females were in poorer physical condition at the time of desertion than were the nondeserting females. This was attributed to the deserter preferentially feeding the nestlings when the quantity of food delivered to the nest by either parent was inadequate to meet the family's requirements. Based on these observations, I developed a dynamic stochastic optimization model to explore the hypotheses that (1) the optimal reproductive strategy of the female Cooper's hawks is the strategy that maximizes her reproductive fitness; and (2) the major factors affecting the female's reproductive fitness and, thus, her optimal strategies are her physical condition, the physical condition of the nestlings, the risks to the nestlings associated with each strategy, and the foraging capabilities of her mate. The model defined reproductive fitness as the weighted average of the nestlings' expected survival and the female's expected reproductive potential in the next breeding season. Using independent sources for the parameters, the model validation compared the model predictions with actual observations of the behavior of radio-tagged, nesting females in this population of Cooper's hawks. The model correctly predicted 90% of the strategies exhibited by the nesting females, which supported the initial hypotheses.