A model is described which aims to predict intake immediately following a change from one food to another that is higher in bulk content; it deals with the transition from one 'equilibrium' intake to another. The system considered is an immature pig fed ad libitum on a single homogeneous food, which is balanced for nutrients and contains no toxins so that the first limiting resource is always energy. It is assumed that an animal has a desired rate of food intake (DFI) which is that needed to meet the energy requirements for protein and lipid deposition and for maintenance. DFI may not be achieved if a bulk constraint to intake exists. Where a bulk constraint operates intake is calculated as constrained food intake (CFI) where (where WHC is the water-holding capacity of the food (kg water/kg dry food) and Cwhc is the animal's capacity for WHC (units/kg live weight per d)). Where intake is not constrained it is assumed that genetic potential will be achieved. Potential growth rate is described by the Gompertz growth function. Where intake is constrained, growth will be less than the potential. Constrained growth rate is predicted as where W is pig weight (kg), EI is energy intake (MJ/d), Em is the energy required for maintenance (MJ/d) and eg is the energy required for unit gain (MJ/kg). The value of eg depends on weight and the fattening characteristics of the pig. Actual growth is predicted to be the lesser of potential and constrained growth. To deal with adaptation it is assumed that the time taken to reach equilibrium depends on the difference in WHC values between the previous and current food and that the capacity to consume food bulk is related to the WHC of the current food. It is proposed that the capacity for WHC on the first day on a new food will be equal to the current capacity for WHC on the last day of the previous food. Thus where FI is food intake (kg/d). Thereafter Cwhc will gradually increase over time to a maximum of 0.27 g/kg. The rate of change in Cwhc is made to be the same for all pigs and all foods. The increase in capacity over time is assumed to be linear at the rate of 0.01 units/d. The model was tested using published data. Qualitatively the predictions of the model were in close agreement with the relevant observed data in at least some cases. It is concluded that the underlying theoretical assumptions of the model are reasonable. However, the model fails to predict initial intake when changed to foods high in wheat-bran content and fails to predict the intake of a non-limiting food where compensatory increases in intake and gain occur. The model could be adapted to overcome the first failure by taking into account the time course of digestive efficiency following a change in food. To deal with the second would require a sufficient understanding of the time course of compensatory growth.