Soon after feeding, fish metabolism increases, resulting in high rates of oxygen consumption (OC) and ammonia excretion (AE). In intensive aquaculture, these peaks must be treated as they occur, requiring water-conditioning equipm ent of high installed capacity. Shaving off the peaks by more frequent feeding reduces the required capacity, but to do that properly, a dynamic prediction model of fish OC and AE is required. Recent experimental OC and AE data from the literature are used to fit a simple, four-compartment (four-state variable) mechanistic digestion-assimilation model to three feeding-frequency treatments. As the AE data are well correlated with the OC data, the dynamic model is used first to predict OC, and then the correlation is used to calculate the corresponding AE. The OC model is composed of a five-parameter, static submodel and a dynamic part with five additional parameters. The latter are fitted with the time-varying data. The resulting fit, to all treatments with the same set of parameters, is good. The dynamic model mimics properly the plateau evident in the once-per-day feeding treatment, as well as the curvatures of the ascending and descending segments. The static portion of the model, based mostly on daily totals, predicts a linear dependence of daily OC and AE on the size of the daily feed ration, in agreement with the data.