1. A model developed over 20 years ago has been used to estimate daily food intake in brown trout living in streams and lakes over a wide geographical range. The chief disadvantages of this early model are that it is not continuous and requires twelve parameters, not all of which can be interpreted biologically. A new model, using a larger data set, was therefore developed to overcome these problems and estimate the mean daily energy intake.
2. The two data sets used to develop the original model were also used to develop the general form of the new one, but a third data set was used to specify the model more precisely and to estimate the parameters. This third data set originated from experiments in which 185 trout (live weight range 1–350 g) were kept individually at 19 constant temperatures (range 3.8–21.7 °C) usually for 5–6 weeks. They were fed freshly killed shrimps (Gammarus pulex) and their food consumption was recorded throughout each experiment.
3. Five, six and eight parameter versions of the new model were all excellent fits to the data (P < 0.001, R2 > 0.99), with the eight parameter version being slightly the best. All parameters can be interpreted in biological terms; three define threshold temperatures, three define the curvilinear slopes in the model over different temperature ranges, one is a weight exponent and one is the maximum daily energy intake of a 1 g trout. The simpler six parameter model was adequate at temperatures above 7 °C.
4. An additional experiment with twenty-eight trout feeding on six different invertebrate foods provided estimates of energy intake that were very similar to those predicted from the model. However, when daily intake was converted to dry weight, agreement with values from the model (also as dry weight) was poor. Possible reasons for this are discussed, as are other studies using the earlier model, and it is shown that different conclusions can be reached depending upon whether comparisons are based on units of energy, dry weight or wet weight.