We conducted two experiments to investigate the extent to which lexical knowledge affects children's interpretation of phonetic input In Experiment one, five-year-old children and adults identified the initial phoneme in stimuli from 18 acoustic continua varying in voice onset time (VOT). For each continuum, one of the two possible phonemic categorizations formed a word (e.g., desk) and the other did not (e.g., tesk). Both children and adults tended to identify the initial phoneme as consistent with a word when the tokens were from the middle of a VOT continuum. In addition, children showed a lexical effect at the endpoints of some continua, where acoustic information was clear. Experiment two addressed one possible explanation of the finding that children, but not adults, are influenced by lexical context even when the acoustic input is clear. Children may adopt a strategy to report words, because of the large number of ambiguous targets (i.e., onsets from within the continua). If so, the lexical effect should disappear when all tokens are clear voiced and voiceless stop consonants from the endpoints of the continua. Five-year-olds identified endpoint stop consonants in words and nonwords. The lexical effect was greatly attenuated for these stimuli relative to the endpoint stimuli in Experiment 1, suggesting a strategic component to children's performance. Nonetheless, a small lexical effect remained for these items. The results suggest that five-year-olds engage in very similar processes to adults enabling then to use lexical context to identify phonemes. Under certain conditions, children weigh lexical context more heavily in making phoneme decisions than do adults.