Modelling is an economic way of assessing pesticide behaviour under field conditions; it is cheaper and faster than field experiments. Modelling attempts to generalize knowledge of pesticide field behaviour through identification of the most important pesticide/soil properties that can be measured in the laboratory. The technology to simulate volatilization of volatile pesticides that are incorporated or injected into the soil is well developed. However, modelling of volatilization rates from plant and soil surfaces before the first significant rainfall event after application is barely possible with current knowledge. The technology to simulate pesticide persistence in the plough layer is well developed; the PERSIST model has been tested at least 178 times, usually resulting in a slightly faster decline in the field than was simulated. In general, available pesticide leaching models are reliable enough to assess the leaching of the bulk of the dose (leaching levels above 1%). The EU drinking water limit of 0.1 μg L1 implies leaching of less than 0.1% of a dose of 1 kg ha1. At such a low leaching level, the validation status of the models is still low, mainly because preferential flow processes in both structured and unstructured soils and the factors controlling the transformation rate in subsoil are not well enough understood.