An action threshold (AT) is one of the most important decision-making elements in integrated pest management. Unlike economic thresholds, ATs are not typically derived from an economic injury level model, but they are more commonly used. ATs may be identified from research-based, pest—crop relationships, but they also may be based on experience. ATs may be adjusted depending on, e.g., weather and plant variety, but modifying ATs to accommodate differences in insecticide efficacy has received little attention. To examine this point, several combinations of ATs and insecticides were evaluated against onion thrips, Thrips tabaci Lindeman (Thysanoptera: Thripidae), a major pest of onion (Allium cepa L.). Studies were conducted in New York onion fields from 2006 to 2008 by using registered insecticides for T. tabaci on onions. We hypothesized that the most efficacious insecticides would provide acceptable control of thrips populations regardless of AT (one, three, and five thrips per leaf), whereas less effective products would only control populations using the lowest AT (one thrips per leaf), Results indicated that T. tabaci infestations were managed effectively when spinetoram was applied after a three larvae per leaf threshold, but not when using lambda-cyhalothrin, methomyl or formetanate hydrochloride. However, T. tabaci infestations were managed well when methomyl and formetanate hydrochloride were applied after a one larva per leaf threshold. T. tabaci infestations were never controlled using lambda-cyhalothrin, regardless of the AT used. None of the products reduced T. tabaci populations to an acceptable level when applied at a five larvae per leaf threshold. Implications of adjusting ATs based on efficacy of different insecticides are discussed.