When forest ecosystems develop over millennia, trees live five centuries, and mortality unfolds over decades, direct repeated observation (hereafter, longitudinal data) may be the only way to understand the fate of forests. Longitudinal data sets contribute greatly to our understanding, complementing experimental, modeling, and chronosequence approaches. Changing climate is changing forests, perhaps most rapidly through altered mortality regimes, and the elusive nature of integrated mechanistic understanding requires refinements and extensions to historically productive protocols. Changing climate reduces the inferential power of chronosequence techniques and changes model parameterization, and only some of the different factors contributing to tree mortality are expected to respond to climate variability and change. Because annual tree mortality rates are 5% to 1% for trees ≥ 1 cm dbh in young and old forests (respectively), detecting changes in mortality rates requires tracking thousands of trees, particularly to examine rare sub-populations of concern (e.g., large-diameter trees). And because mortality factors can be spatially aggregated and density-dependent, the causes and rates of tree mortality depend on the relationships between forest spatial structure and the direct and indirect effects of climate. Permanent plots with a combination of larger size, higher spatial precision, and greater sampling frequency will be required to further elucidate spatially explicit aspects of western forest demography. The combination of the longitudinal protocols developed by the Smithsonian Center for Tropical Forest Science, originally for studying tropical forest species diversity, and those developed by the US Geological Survey for annual tree mortality assessment together uniquely allow robust investigation of climate-mediated change in temperate forests.