To determine the relative effectiveness of two different types of exposure on young children's preference for initially novel foods, 51 two- to five-year-old children received either "look" or "taste" exposures to seven novel fruits. Foods were exposed five, 10 or 15 times, and one food remained novel. Following the exposures, children made two judgments of each of the 21 pairs: one based on looking, the other on tasting the foods. Thurstone Case V scaling solutions were correlated with exposure frequency, and these were significant for the visual judgments of the looked at foods (r = 0.91), the visual judgments of the tasted (and looked at) foods (r = 0.97) and the taste judgments of the tasted foods (r = 0.94). The only non-significant relationship was for the taste judgments of the looked at (but never tasted) foods (r = 0.24). The results indicate that to obtain significant positive changes in preference, experience with the food must include experience in the modality that is relevant for the judgments. While visual experience produced enhanced visual preference judgments, visual experience was not sufficient to produce significantly enhanced taste preferences. This finding is consistent with a "learned safety" interpretation of the exposure effects noted in the taste judgments: experiences with novel tastes that are not followed by negative gastrointestinal consequences can produce enhanced taste preference.