Variability is a key feature of the pelagic ecosystems of the Southern Ocean and an important aspect of the variation is fluctuation in the abundance of krill, the major prey item of many of the higher predators. Direct impacts of variability in the large-scale physical environment, such as changes in ocean circulation, have been suggested as the main factor generating the observed fluctuations. So far, however, there has been little quantitative assessment of the importance of krill population dynamics in the observed variation. Here, analyses of a model of krill population development and predator diet data from South Georgia were used to examine seasonal changes in the population structure of krill. The krill population model was combined with a size-based selection function and used to generate expected length-frequency distributions in the predator diet through a summer season. Comparison of the model solutions with the predator diet data indicates that the model can reproduce the observed pattern of variation and emphasizes that adult population changes are a key aspect of the interannual fluctuations observed during some years. Low krill abundance was associated with reduced representation of the 3+ age group, whereas when krill were abundant the 3+ age class was the major age group present. The seasonal changes in the population structure in the predator diet involve a complex interaction of relative year class strength, timing of immigration, fluctuations in growth rates and dynamic predator selective effects. Development of the model to examine the interactive effects of changing krill growth and mortality rates will be a valuable next step. The dominance of the changes in krill population age structure underlines the fact that to understand the variability of the South Georgia ecosystem we must identify the major factors generating variability in population dynamics throughout the Scotia Sea.