Modeling fish growth to predict fish production is important for planning of aquaculture development. Since growth strongly depends on feed intake, a growth model should be able to predict maximum feed intake of the fish under ad libitum feeding. However, the equations for maximum feed intake in the existing fish growth models are still descriptive and do not reflect the underlying mechanisms regulating feed intake. Feed intake in fish is regulated by a number of factors. Physiologically, effects of metabolites in the blood are very important. The appearance of the metabolites in the blood at a rate greater than that at which they are removed signals satiety. Environmentally, oxygen may be a major determinant of maximum feed intake. The uptake of oxygen for metabolism is limited by the gill surface area and dissolved oxygen concentration in the water. When oxygen supply does not satisfy oxygen demand, fish may stop eating. Oxygen supply and the accumulation of metabolites in the cells and the blood are linked. In any case, inclusion of pools of metabolites in growth models is essential to simulate feed intake regulation. This study aimed to develop a dynamic explanatory fish growth model which incorporates pools of metabolites as potential regulators of metabolism and feed intake. Initially, we considered only the effect of glucose on feed intake, and parameterized and calibrated the model for rainbow trout (Oncorhynchus mykiss Walbaum), as more information related to this species was available. State variables in the model are pools of metabolites and body constituents, including amino acids, fatty acids, glucose, acetyl CoA and body protein and fat. It was assumed that the conversions between the state variables are influenced by the concentration of the metabolites involved. Concepts of enzyme kinetics were adapted to formulate the equations representing the conversion rates. Feed intake was modeled based on the glucose static theory, which states that the satiety center in the brain is stimulated by an increase of glucose in the blood, causing a reduction in feed intake. In the model, when glucose concentration is higher than a threshold, fish cease to eat until glucose concentration drops below that point. Growth of the fish was calculated based on the relationship between body weight and protein biomass. Estimation of the parameters in the rate equations was done indirectly based on general principles of nutrient metabolism and experiments on nutritional physiology of rainbow trout. After parameterization, the model was calibrated with experimental data on growth of rainbow trout under restricted feeding. A computer program was developed in Delphi 7 for the simulations, where the differential equations were solved numerically using the Euler method with a fixed time step of 0.01 day. Agreement between the simulated and experimental fish weight was assessed based on the magnitude of the relative errors (RE) and the average relative error (ARE), which were calculated as: n SWi − EWi RE i = 100 × and ARE = ∑ RE i12 (SWi + EWi ) i =1 where REi is the relative error for case i, SWi and EWi are the simulated and observed fresh weight of the fish for case i, respectively. The model predicted fresh weight of the fish with an average relative error of -0.8% (range -18.2% to 17.8%). The parameterization and simulation results showed the gaps in our knowledge about the real values of maximum conversion rates of metabolites and the conditions under which these maximum rates occur. Results of experiments in which the fish are fed ad libitum with feed containing various levels of carbohydrates are needed for calibration and validation of the module for feed intake simulation.