Attempts to gauge the biological impacts of climate change have typically focussed on the lower levels of organization (individuals to populations), rather than considering more complex multi-species systems, such as entire ecological networks (food webs, mutualistic and host-parasitoid networks). We evaluate the possibility that a few principal drivers underpin network-level responses to climate change, and that these drivers can be studied to develop a more coherent theoretical framework than is currently provided by phenomenological approaches. For instance, warming will elevate individual ectotherm metabolic rates, and direct and indirect effects of changes in atmospheric conditions are expected to alter the stoichiometry of interactions between primary consumers and basal resources; these effects are general and pervasive, and will permeate through the entire networks that they affect. In addition, changes in the density and viscosity of aqueous media could alter interactions among very small organisms and disrupt the pycnoclines that currently compartmentalize many aquatic networks in time and space. We identify a range of approaches and potential model systems that are particularly well suited to network-level studies within the context of climate change. We also highlight potentially fruitful areas of research with a view to improving our predictive power regarding climate change impacts on networks. We focus throughout on mechanistic approaches rooted in first principles that demonstrate potential for application across a wide range of taxa and systems.