A recent International Civil Aviation Organization initiative mandates that pilots and air-traffic controllers operating on international routes demonstrate adequate English-language proficiency for successful communication. The Versant Aviation English Test was developed to serve this purpose. It is a fully automated speaking and listening performance test, where administration of the test tasks and scoring of the candidates' responses are computerized. We argue that not only do candidates engage in cognitively and linguistically appropriate interactions, but that computer-generated scores and human ratings are consistent (r = 0.94) , enabling valid score-based decisions to be made on the basis of automated language testing.