In pesticide risk assessment, effect concentrations and dynamics of long-term community-level effects caused by pulse exposures remain to be investigated. This is because long-term experiments are exceptionally rare, and most of the previously investigated communities had low proportions of sensitive long-living species. The aim of the present study was to investigate the effect of a single pulse contamination with the insecticide thiacloprid on invertebrates. We employed mesocosms designed to realistically mimic communities in small streams within the agricultural landscape. Specifically, the objectives were to (i) compare the community Lowest-Observed-Effect Concentration (LOEC) with organism-level median lethal concentrations (LC50), and (ii) to assess recovery dynamics with special focus on short- and long-living taxa. The contamination resulted in long-term alteration of the overall invertebrate community structure (7 months, until the end of the experiment). Long-term community LOEC was 3.2 μg/L (Redundancy Analysis), slightly below the acute LC50s known for sensitive invertebrates relevant to the mesocosm community. However, one species (stonefly Nemoura cinerea) was affected at the lowest tested concentration, 70 times below the lowest known LC50. Concerning time to recovery from the effect, we found that the duration depends on the life-cycle characteristics of species, but not on the toxicant concentration: short-living (mulivoltine) species recovered after 10 weeks following contamination, whereas long-living (uni- and semivoltine) species did not recover until the end of the experiment (7 months). The present example shows that concentrations of pesticides at which majority of the species is affected can be predicted by acute organism-level toxicity tests with sensitive species. However, tests with longer observation periods, as well as consideration of environmental factors and inter-taxon variability in sensitivity are required to predict effects on all species comprising a community. Realistic prediction of community recovery dynamics requires consideration of the species' life-cycle traits.