The link between pelagic and benthic systems has been suggested to be particularly strong on Arctic continental shelves. Differing sampling strategies and the multitude of pelagic and benthic parameters that have been compared, however, make it difficult to determine the limitations of this important ecosystem concept. We sampled across multiple gradients in water mass characteristics, ice cover, productivity regimes, and seasons to test the generality of the pelagic–benthic coupling paradigm during a 3-year study (2003–2005) in the Barents Sea. While benthic community structure varied among stations, biomass was not significantly correlated with sediment oxygen demand (SOD), a finding contrary to some published studies. Average SOD varied from 2.3 to 7.3mmol O2m−2d−1, and was (positively) correlated as well with sediment pigments, as it was with the vertical flux of pigments and carbon measured by sediment traps deployed at 90m. Export flux was also strongly coupled with our measure of benthic carbon cycling (SOD), suggesting that both processes vary on similar, short time scales (days to weeks). In contrast, processes responsible for patterns in benthic biomass reflect a response over longer time scales (several years), and are thus not well coupled with SOD. We conclude that conflicting results of studies assessing pelagic–benthic coupling may not necessarily be due to general inconsistencies in the fundamental relationship, but instead to improper comparisons of parameters or processes that vary over different scales. These findings are particularly relevant for evaluating system processes in ecosystem modeling studies.