Variation in demographic rates within a spatially structured population could have important consequences for management decisions, harvesting strategies and offtake rates. Although there is a growing body of evidence suggesting that demographic rates vary within populations over a range of spatial scales, there has been little research investigating the consequences of this variation for population management. In this paper, data on the dynamics of two female red deer sub-populations on Rum are analysed, and evidence is presented for differences between the fecundity and mortality rates of the two sub-populations. A simple harvesting model is developed to represent the dynamics of the two sub-populations, including density-independent migration between sub-populations and spatially correlated environmental variability. The highest monetary yield in the model is obtained by harvesting the more resilient sub-population at a higher rate. Surprisingly the losses involved in harvesting both sub-populations at the same rate are insignificant. However, if migration were density-dependent, the size of one sub-population would be more relevant to harvesting policy for the other sub-population. The results of this empirical study are compared to theoretical work on spatially structured populations; it is shown that when a species has complex age- and sex-structured population dynamics, previous theoretical results may not hold.