The key to understanding the link between climate variability and the biosphere are feedbacks of the ocean and land surfaces to atmospheric variability. Generalizations of the Hasselman random forcing model continue to provide a framework for much of the observed climate variability. Zero dimension ocean-atmosphere and land-atmosphere coupled models reveal the timescales of mutual adjustment and hence the amplitude of variability for a given forcing. The coupling in both models introduces longer timescales than in any uncoupled model and hence illustrates how the coupled systems act to amplify climate variability. The common feature of both models is a quantity that is conserved between the surface and the atmosphere. For the atmosphere-ocean model, this is energy, whereas for the atmosphere-land model, the conserved quantity is water. These conservative quantities represent the projection of climate variability onto the slowest system mode, an inference that readily generalizes to modes of climate variability with spatial structure and hence they provide a useful focus for diagnosis of observations and complex model simulations. ¿ 2000 American Geophysical Union