Understanding the temporal variability in vital rates (e.g. growth and survivorship) of wild populations is practically and statistically difficult but crucial for connecting such variation to mechanistic drivers and their population consequences. For somatic growth, empirical estimates of variation are rare because they often require expensive long-term tag-recapture programs. In marine pelagic ecosystems, where many species are difficult to sample and not amenable to tagging studies, researchers have relied on estimating growth from length-frequency distributions. We developed a general approach for estimating growth from observed length-frequency samples by combining kernel density estimates of the length-frequency distribution and the von Bertalanffy growth function. Our approach is conceptually straightforward and easy to implement. We applied the methods to Antarctic krill Euphausia superba collected from the Southern Ocean over a span of 19 yr to document among-year variation in krill growth during the austral summer. Our estimates of growth align closely with existing estimates of growth, but we provide the first estimates of among-year variation in krill growth. We estimate very high among-year variation in growth (annual estimates for a 30 mm krill in the Elephant Island region ranged from 0.00 to0.17 mm d −1 ; mean = 0.073, among-year coefficient of variation ≈ 0.8). We correlated growth rate variation to estimates of ocean chlorophyll but not to other oceanographic indices, contrasting with results from previous studies. The large amount of variation in growth unexplained by environmental covariates has substantial implications for ecosystem management in the Southern Ocean ecosystem.