Water resource managers require catchment-scale nutrient transport models t
o help in the investigation and control of diffuse pollution. Many differen
t models exist for water and pollutant movement, causing problems of select
ion because each model has often been constructed for specific purposes, an
d the choice of an appropriate one is difficult: wrongly applied models can
lead to unreliable results and misleading information. The selection of a
model for estimating phosphorus loss from a lowland English catchment is de
scribed. 14 models, whose underlying principles had been published, were ev
aluated. 'SWAT' was found to be the most suitable. It has several favourabl
e qualities, but also several shortfalls which may impair its predictive ab
ility. Their identification enables the model code to be re-written. Reliab
ility estimation is needed to ensure that the ranges of model prediction ar
e known.