Background/Question/Methods Forest stands undergo long-term changes in composition, structure, and function as a consequence of: internal processes related to demography and competitive interactions; disturbances at a variety of scales, frequencies and causes, and changes in basic environmental conditions, such as climate.  The long time periods involved make the quantitative study of many aspects of forest development difficult.  As a consequence, there is much reliance on short-term observations, studies of chronosequences, and modeling in the development and testing ecological theory.  There also seems to be an unfortunate tendency for model outputs to be accepted as real, rather than viewed as working hypotheses.

Results/Conclusions Ultimately real-world data are necessary to critically test ecological theory related to forest tree demography and long-term changes in the structure and function of ecosystems.  Long-term observations of large forest plots are the primary source for these data.  Such plots not only allow us to critically test theory but also stimulate the development of new theoretical constructs.  They can also provide early warming systems for important but subtle changes in demography that are occurring as a result of climate change. A global system of permanent sample plots in natural forest ecosystems is needed that will provide representative data on demographic processes, with particular emphasis on rates, causes, and patterns of mortality, and on key structural and functional attributes of these ecosystems.  A forest plot network might well be linked to the National Ecological Observatory (NEON) program in the United States.  Challenges and important attributes of such a plot network will be discussed, using examples drawn from our experiences in the forests of the Pacific Northwest.