When people switch between languages, inhibition of currently irrelevant languages is assumed to occur. The authors examined inhibition of irrelevant languages with a cued language-switching paradigm. A cue indicated in which of 3 languages (German, English, or French) a visual stimulus was to be named. In 2 experiments, the authors found that naming latencies were increased in n-2 language repetitions (e.g., German/English/German) compared with in n-2 language nonrepetitions (e.g., French/English/German). This difference ("n-2 repetition costs") indicates persisting inhibition of abandoned languages. It is important to note that n-2 language-repetition costs also occurred in conditions in which the language but not the cue (Experiment 1) or the stimulus/response set (Experiment 2) repeated from trial n-2 to trial n. These data demonstrate that inhibition is not restricted to a specific cue or stimulus/response set. Rather, the data suggest more global inhibitory processes that affect the mental representation of competing languages. (Contains 3 tables.)