Ecotoxicological field studies have shown that total abundance and biomass often recover shortly after pulsed toxicant stress. In contrast, population structure showed comparatively long-term alterations before reaching pre-treatment conditions. We investigated two mechanisms that may explain the prolonged recovery of population structure: latent toxicant effects on life-history traits on the individual level and competition on the population level. To test these hypotheses we exposed populations of Daphnia magna to a pulse of the pyrethroid Fenvalerate. For several generations the populations were kept at two different degrees of competition: strong competition at carrying capacity and reduced competition maintained by simulated predation. After disturbance due to Fenvalerate exposure, biomass recovered after 14–17 days. In contrast, size structure characterised by a lack of large and dominance of small organisms recovered after 43 days in populations with strong competition. Size structure recovered twice faster in populations with reduced competition. We explain this as follows: due to toxicant induced mortality, food availability and consequently birth rate increased and populations were dominated by small individuals. In populations without predation, these cohorts grew and eventually exerted high intraspecific competition that (i) stopped further growth of juveniles and (ii) increased mortality of adults. These demographic processes were mainly responsible for the prolonged recovery of size structure. In contrast, for populations with predation, the regular harvest of individuals reduced competition. Juveniles developed continuously, allowing a fast recovery of size structure in these dynamic populations. In risk assessment the duration for populations to recover from (toxicant) stress, is crucial for the determination of ecological acceptable effects. We conclude that competition needs to be considered in order to understand and predict recovery of size structure.