The growing awareness about the need to anticipate the future of land systems focuses on how well we understand the interactions between society and environmental processes within a complexity framework. A major barrier to understanding is insufficient attention given to long (multidecadal) temporal perspectives on complex system behavior that can provide insights through both analog and evolutionary approaches. Analogs are useful in generating typologies of generic system behavior, whereas evolutionary assessments provide insight into site-specific system properties. Four dimensions of these properties: (1) trends and trajectories, (2) frequencies, thresholds and alternate steady states, (3) slow and fast processes, and (4) legacies and contingencies, are discussed. Compilations and analyses of past information and data from instruments and observations, palaeoenvironmental archives, and human and environmental history are now the subject of major international effort. The embedding of empirical information over multidecadal timescales in attempts to define and model sustainable and adaptive management of land systems is now not only possible, but also necessary.