Breeding success in wild mallards was studied on small eutrophic nemoral lakes in a two-year cross-over experiment in which wing-clipped conspecifics were added to increase pair density. The number of wild mallards that came to the lakes to nest (prior to introductions) did not differ between years. Introduction treatment led to a significant reduction in brood number in wild mallards, but it did not alter the number of stage 2+ ducklings finally produced on a lake. Introduction had no effect on lake utilization by broods, ducklings and non-breeding adults (cumulative days over the entire breeding season). Abundance of invertebrate prey differed greatly among lakes, but it did not correlate with breeding success. Breeding success was thus subject to sequential density-dependence; i.e. a lower number of broods still produced the same number of 2+ ducklings. We speculate that predation is the most likely process behind both patterns. We conclude that late and snapshot measures of duckling productivity may mask density dependent population processes of fundamental importance to regulation and harvest policy.