Objective: To compare children's food preferences longitudinally and identify factors related to food preferences.

Design: Mothers completed the Food Preference Questionnaire for children at 2 to 3 years of age (T1), 4 years (T2), and 8 years (T3) and for themselves at T1 and T3. Both groups completed a Food Neophobia Scale at T3.

Subjects: 70 child/mother pairs who had participated continuously in the longitudinal study.

Statistical analyses performed: Changes in food preferences over time were tested with paired t tests and correlations. Consistency percentages were calculated by summing the consistent matches (like/like) for each food between two time periods. Similarly, concordance percentages were calculated for child/mother pairs by summing the concordant matches for each food. General linear models were developed to identify influences on children's food preferences.

Results: Although children liked most foods, the number of liked foods did not change significantly during the 5 to 5.7 years of the study. The strongest predictors of the number of foods liked at age 8 years (R2=0.74) were the number liked at 4 years (P<.0001) and the food neophobia score (P=.0003). Newly tasted foods were more likely to be accepted between T1 and T2 than T2 and T3. Mothers' and children's food preferences were significantly but moderately related. Foods disliked by mothers tended not to be offered to children.

Applications/conclusions: The important role of children's early food preferences is confirmed by this study. Mothers influence children via their own preferences, which may limit foods offered to children.