Climate-land surface interactions have been mainly investigated either in terms of short-term modification of fluxes of matter, momentum, and energy, which primarily affect the atmosphere, or in the long-term, biogeographical impacts of climatic conditions on the type of vegetation that occupies a site. Logically, there must be time and space scales at which these short-term and biogeographical climate-land surface interactions are both relevant. It is proposed that it is these intermediate scales, at which physiological and biogeographical processes cannot easily be separated, which are most relevant in the study of climate change, and that new analytical and modeling approaches are needed which include both. Moreover, periods of rapid climate change and periods of increasing anthropogenic impacts on the land surface can be expected to induce a wide range of transient vegetation dynamics. The timescale of interest here excludes equilibria and demands a consideration of the outcomes of processes that have a wide variety of temporal frequencies. A recent workshop brought together ecologists and climatologists with interests in observing, modeling, and predicting the dynamics of land-atmosphere processes at the interannual to decadal timescale. The result was a proposal to exploit recently developed archives of remotely sensed data and others such as paleobotanical and paleoclimatological records in order to advance beyond the polarized concept of land-atmosphere processes which comes from a consideration only of short and long time periods, while ignoring their interactions. The new demands of this agenda for observational and analytical methods are considered. ¿ 2000 American Geophysical Union