Control failure of pests and selectivity of insecticides to beneficial arthropods are key data for the implementation of Integrated Pest Management (IPM) programs. Therefore, the aim of this study was to assess the control failure likelihood of Plutella xylostella and the physiological selectivity active ingredients to parasitoid Oomyzus sokolowskii (Hymenoptera: Eulophidae) and to predators Polybia scutellaris (Hymenoptera: Vespidae) and Lasiochilus sp. (Hemiptera: Anthocoridae). In bioassays, P. xylostella larvae and O. sokolowskii, P. scutellaris and Lasiochilus sp. adults were used. Concentration-mortality curves of six insecticides for P. xylostella were established. These curves were used to estimate the mortality of P. xylostella at the recommended concentration, in order to check a control failure of insecticides to this pest. Furthermore, the lethal concentration for 90% of populations (LC90) and the half of LC90 were used in bioassays with the natural enemies to determine the selectivity of these insects to insecticides. All tested insecticides showed control failure to P. xylostella, indicated by high LC90 and low estimated mortalities (less than 80%). The cartap insecticide was selective in half of LC90 to Lasiochilus sp. and moderately selective in LC90 and the half of LC90, to Lasiochilus sp. and P. scutellaris, respectively. Deltamethrin was moderately selective in the half of LC90 to predator Lasiochilus sp. Cartap, carbaryl, and deltamethrin reduced the mortality of Lasiochilus sp. in the half LC90. The results also showed that the insecticides methamidophos, carbaryl, parathion methyl and permethrin were not selective to any of the tested natural enemies. The role of insecticides in IPM systems of Brassica crops is discussed based on their control failures to P. xylostella and selectivity to their natural enemies.