A comparison was made of a range of different models of wool pesticide breakdown. Linear models, with a constant rate of breakdown as the wool grows, were always inferior to models that allowed the breakdown rate to decrease as the length of wool increased. Two-pool models were usually satisfactory and only minor or no improvement was gained by inclusion of additional pools or more complex models. There was little difference in fit between models with two-way diffusion compared with one-way transfer. The ‘standard’ model, reported previously, uses two exponential functions rather than multiple pools and was usually the best fit or at least as good a fit as the two- or three-pool models. We recommend that this model continue to be used, although the two-pool model is adequate and may be easier to implement and visualise. However, the three-pool model without transfer is also satisfactory and much faster to evaluate.