Abstract. Over time scales of thousands to millions of years, and in the absence of rejuvenating disturbances that initiate primary or early secondary succession, ecosystem properties such as net primary productivity, decomposition, and rates of nutrient cycling undergo substantial declines termed ecosystem retrogression. Retrogression results from the depletion or reduction in the availability of nutrients, and can only be reversed through rejuvenating disturbance that resets the system; this differs from age-related declines in forest productivity that are driven by shorter-term depression of nutrient availability and plant ecophysiological process rates that occur during succession. Here we review and synthesize the findings from studies of long-term chronosequences that include retrogressive stages for systems spanning the boreal, temperate, and subtropical zones. Ecosystem retrogression has been described by ecologists, biogeochemists, geologists, and pedologists, each of which has developed somewhat independent conceptual frameworks; our review seeks to unify this literature in order to better understand the causes and consequences of retrogression. Studies of retrogression have improved our knowledge of how long-term pedogenic changes drive shorter-term biological processes, as well as the consequences of these changes for ecosystem development. Our synthesis also reveals that similar patterns of retrogression (involving reduced soil fertility, predictable shifts in organismic traits, and ecological processes) occur in systems with vastly different climatic regimes, geologic substrates, and vegetation types, even though the timescales and mechanisms driving retrogression may vary greatly among sites. Studies on retrogression also provide evidence that in many regions, high biomass or "climax" forests are often transient, and do not persist indefinitely in the absence of rejuvenating disturbance. Finally, our review highlights that studies on retrogressive chronosequences in contrasting regions provide unparalleled opportunities for developing general principles about the long-term feedbacks between biological communities and pedogenic processes, and how these control ecosystem development.