The biological carbon pump is responsible for the transport of similar to 5-20 Pg C yr(-1) from the surface into the deep ocean but its variability is poorly understood due to an incomplete mechanistic understanding of the complex underlying planktonic processes. In fact, algorithms designed to estimate carbon export from satellite products incorporate fundamentally different assumptions about the relationships between plankton biomass, productivity, and export efficiency. To test the alternate formulations of export efficiency in remote-sensing algorithms formulated by Dunne et al. (2005), Laws et al. (2011), Henson et al. (2011), and Siegel et al. (2014), we have compiled in situ measurements (temperature, chlorophyll, primary production, phytoplankton biomass and size structure, grazing rates, net chlorophyll change, and carbon export) made during Lagrangian process studies on seven cruises in the California Current Ecosystem and Costa Rica Dome. A food-web based approach formulated by Siegel et al. (2014) performs as well or better than other empirical formulations, while simultaneously providing reasonable estimates of protozoan and mesozooplankton grazing rates. By tuning the Siegel et al. (2014) algorithm to match in situ grazing rates more accurately, we also obtain better in situ carbon export measurements. Adequate representations of food-web relationships and grazing dynamics are therefore crucial to improving the accuracy of export predictions made from satellite-derived products. Nevertheless, considerable unexplained variance in export remains and must be explored before we can reliably use remote sensing products to assess the impact of climate change on biologically mediated carbon sequestration.