This article discusses the implications of "multicompetence" (Cook, 1991), defined as an individual's knowledge of more than one language, for language assessment. The growing psycholinguistic evidence highlighting variability in native speaker performance as a result of multicompetence is reviewed, evidence that further questions the validity of the much beleaguered notion of the "native speaker benchmark" when such benchmarks are based on monolingual native speakers. The possibility of multicompetent benchmarks is considered, and preliminary descriptions of a variety of potential linguistic profiles are provided, delineated by linguistic subsystem and language combination, along with suggestions for further work in this area. Recognizing that research on multicompetence is still in its infancy, and that adoption of partial descriptions of multicompetent grammars for the purposes of language testing poses a number of logistical challenges, the article concludes with some general suggestions for standardized assessments, and assessments in foreign and second language contexts. These recommendations relate to several themes in Applied Linguistics, including proposals currently under consideration for the assessment of English as a Lingua Franca and World Englishes. (Contains 3 footnotes.)