Climates, ecosystems, and societies interact over a tremendous range of temporal and spatial scales. Scholarly work on climate impacts has tended to emphasize ~fferent questions, variables, and modes of explanation depending on the primary scale of interest. Much of the current debate on cause and effect, vulnerability, marginality, and the like stems from uncritical or unconscious efforts to transfer experience, eonciusions, and insights across scales. This paper sketches a perspective from which the relative temporal and spatial dimensions of climatic, ecological, and social processes can be more clearly perceived, and their potential interactions more critically evaluated. Quantitative estimates of a variety of charaeterisLic scales are derived and compared, leading to specific recommendations for the design of climate impact studies.