Although virtually all climate records with annual resolution record decadal-centennial scale climate fluctuations, their origin is obscure. Three candidates have been proposed-solar variability, volcanism and internal variability in the ocean-atmosphere system. A brief review of available evidence suggests that there is some support for the influence of solar variability, that the importance of volcanism may have been overestimated and that there is at present only limited information on the thermohaline circulation that suggests perhaps at most a 10–20% variation around its mean Holocene value. It is suggested that each of the above mechanism has a different geographic signature of response in the surface temperature field. Some examples from energy balance model calculations indicate that: (1) solar variability has a near-global response, with the amplitude of response slightly larger over land; (2) volcanism has a proportionately larger amplitude of response over land than over ocean; and (3) the most oft-cited mode of internal variability, changes in the North Atlantic thermohaline circulation, has a hemispheric asymmetry in response. Preliminary comparison of solar results with observations indicates that, if the solar influence exists, it is not being manifested in terms of simple cooling; changes in the ocean-atmosphere system may be significantly modifying the response. Despite this disagreement it is proposed that climate model results can nevertheless be used to postulate key ‘centers of action’ that should be most sensitive to partitioning the effects of the above mechanisms. One example of a sampling strategy is presented. Although this strategy can be imporved by further analysis of data and model calculations, it is hoped that it will contribute to imposing a