Surplus production rates predicted by simple biomass dynamics models are generally expected to follow a simple dome-shaped pattern as population size changes and to show similar trajectories during population decline and recovery. Age-structured models, however, predict substantially lower surplus production rates during population recovery than during decline because of reduced mean fecundity, unless recruitment compensation is very strong. Ecosystem models like Ecosim predict more complex patterns, with reduced production during recoveries due to both age-structure effects and cultivation–depensation effects related to changes in competitor and predator abundances. Production-driven recoveries, where surplus production per biomass is higher during recovery than decline, are predicted in cases where there has been substantial change in overall ecosystem productivity or community structure. 110 case examples illustrate that simple, repeatable relationships between stock size and production are uncommon, and the most common pattern is production-driven change in stock size, where changes in production rate apparently independent of stock size then drive stock increase or decrease. We conclude that nonstationarity in productivity needs to be considered as part of population rebuilding and that empirical estimates of surplus production may provide insight in this process.