Five experiments were conducted to investigate psychological determinants of food preference in five to seven year-old children. The research was informed by: (i) the general literature on human food preference, and (ii) behaviour analytic theory and research, particularly that on rule-governance. Experiment 1, using a between groups design, examined the impact of in vivo peer behaviour on novel food consumption. Children exposed to "positive" peers avidly consuming a target food tended to prefer that food. Conversely, children who observed "negative" peers, rejecting the food, consumed little: it was also shown that these negative effects could be largely overridden by subsequent exposure to positive peers. Experiments 2 to 5 utilised multiple baseline designs to evaluate the effectiveness of a series of multi-component interventions designed to promote consumption of previously refused fruits and vegetables. To maximise ecological validity and long-term maintenance of behaviour change these experiments were conducted in subjects' homes in the context of the evening meal. During Experiment 2 an intervention incorporating video modelling, contingent rewards, and instructions effectively promoted consumption of three named foods. However, little generalisation to the consumption of other foods was evident. In Experiment 3 a similar intervention targeted broader food categories (vegetables and fruit) and this was effective in promoting consumption of up to 12 foods. Maintenance interventions utilising token rewards were effective in promoting long term consumption. During Experiment 4, written instructions and contingent rewards, without video modelling, were relatively effective in promoting fruit, but not vegetable consumption. During Experiment 5, instructions and video modelling, without contingent rewards had a negligible effect on the consumption of either food category. The results demonstrated that, contrary to the widely held belief within the human food preference literature, interventions utilising contingent rewards can be very effective in modifying food preferences. In discussing the results consideration was given to: (i) the role of rulegovernance in ensuring effective reward use; (ii) maintenance and generalisation of behaviour change; and (iii) the outcome measure most appropriate for food preference research.