Mortality of agricultural pests caused by arthropod predators is a valuable ecosystem service for crop production. The earwig, Euborellia annulipes (Lucas), attacks different pest species in various crop ecosystems, including larvae and pupae of the boll weevil, Anthonomus grandis grandis (Boh.). Despite such biological control, cotton pest management remains heavily dependent on synthetic insecticides. In this study, multiple factors were assessed to measure the selectivity of insecticides used against sapsucking and chewing cotton pests for two E. annulipes populations. Nymphs and adults of E. annulipes were exposed to the insecticides in two ways: ingestion of contaminated prey, and contact with dried residues on either inert surfaces or treated plants bearing prey. Pymetrozine, chlorantraniliprole, and spinetoram had little effect on the predator regardless the tested earwig population, life stage, or the route of exposure. Cyantraniliprole affected the predator in some life stages and through some types of contact. Pyriproxyfen was harmless to adult earwigs, but prevented normal development of nymphs to adults. Chlorfenapyr, indoxacarb, lambda-cyhalothrin, chlorpyrifos, dimethoate, and malathion were harmful to the predator regardless earwig life stage or method of exposure. The negative impact was diminished when exposure occurred on plants with predator allowed to shelter in the soil. The results indicate that insecticide selectivity outcome is a multi-factor driven by the insecticide, predator life stage and the redator's behavior. Therefore, testing different predator life stages via several routes of exposure, without denying the insect the opportunity to engage in its normal behaviors can provide better estimates of insecticide selectivity.