This paper evaluates and generalises state-of-the-art approaches for dynamic modelling of bioaccumulation in fish resulting from short duration liquid discharges of radionuclides (3H, 14C, 60Co, 134Cs, 137Cs, 65Zn, 89Sr, 90Sr, 125I, 131I, 241Am, isotopes of Pu and U) to rivers. Based on a review of model parameter values, predictions are made of maximum and timeintegrated activity concentrations in fish. A simplified version of the model was developed and presented as look-up graphs. The influence of various environmental parameters on model output was evaluated by sensitivity analysis. Maximum and time-integrated concentrations in fish may be predicted for rivers based on the river volumetric flow rate and water temperature. It is demonstrated that the dynamic model gives lower and more realistic predictions of maximum concentrations in fish than the simpler Concentration Factor approach. However, for time-integrated concentration in fish, and estimation of radiation dose to humans from consumption of the fish, the Concentration Factor approach gives similar predictions to the dynamic model.