Objective: It was examined whether caloric conditioning or social learning strategies dominate in taste preference acquisition in children. The caloric learning paradigm predicts that eating or drinking artificially sweetened products, which deliver virtually no energy, will not lead to a taste preference whereas the social learning paradigm predicts that seeing important others modelling the eating and drinking of these 'light' products will induce a preference for the taste of light products in the child. Design: In a 2x2 between subjects factorial design, the amount of energy and social modelling was varied. Setting: The study was undertaken at primary schools in Maastricht, The Netherlands. Subjects: Forty-five children participated and six children dropped out. The 39 children who completed the study (14 boys and 25 girls) had a mean age of 67 months (range 51-81, s.d. 5.6). Interventions: Each subject took part in nine conditioning trials with an individually selected tasting yoghurt which was not preferred very much at the pre-test. Results: The children in the combined caloric and social condition showed an increase in their preference for the conditioned taste which was larger than a regression-to-the-mean effect (P = 0.007), whereas children in the other groups did not. Conclusion: Caloric and social learning combined, ie modelling the consumption of energy-rich foods or drinks, is the best way to establish taste preferences. Children more easily learn a preference for energy-rich food that is eaten by significant others than for food that is low in energy and eaten by significant others.