Previous research has shown that experimenter-presented masculine generics can create male bias in the gender content of subjects' imagery. The present study tests experimentally whether subjects' own use of masculine generics has a similar effect on their imagery. College student subjects were induced to complete sentence fragments using masculine or unbiased generics, then asked to describe their imagery for each sentence and to give a first name to fit the person they visualized for each sentence. These dependent measures were coded for gender, and as predicted, analysis of variance showed that male bias was higher in the masculine generic condition than in the unbiased condition. Also as predicted, male subjects were more male-biased overall than were female subjects. The findings are discussed in terms of linguistic relativity (the proposition that language can shape thought), prototypicality (the most typical he is probably a man), and activation of multiple meanings (he has male-specific and gender-neutral denotations, and both may be activated even when the gender-neutral meaning is intended).