Research on the consequences of climate change on agriculture, forestry and fisheries is addressing deepening levels of system complexity that require a new suite of methodologies to cope with the added uncertainty that accompanies the addition of new, often non-linear, process knowledge. The added realism of experiments (e.g., FACE) and the translation of experimental results to process crop-simulation models are adding confidence to model estimates. Integrated physiological and economic models (e.g., Fischer et al., 2005a) allow holistic simulation of climate change effects on agricultural productivity, input and output prices, and risk of hunger in specific regions, although these simulations rely on a small set of component models. The application of meta-analysis to agriculture, forestry and fisheries in order to identify trends and consistent findings across large numbers of studies has revealed important new information since the TAR, especially on the direct effects of atmospheric CO2 on crop and forest productivity (e.g., Ainsworth and Long, 2005) and fisheries (Allison et al., 2005). The complexity of processes that determine adaptive capacity dictates an increasing regional focus to studies in order best to understand and predict adaptive processes (Kates and Wilbanks, 2003): hence the rise in numbers of regional-scale studies. This increases the need for more robust methods to scale local findings to larger regions, such as the use of multi-level modelling (Easterling and Polsky, 2004). Further complexity is contributed by the growing number of scenarios of future climate and society that drive inputs to the models (Naki?enovi? and Swart, 2000).