Single domain models may seriously underestimate leaching of nutrients and pesticides to groundwater in clay soils with shrinkage cracks. Various two-domain models have been developed, either empirical or physically based, which take into account the eects of cracks on water ¯ow and solute transport. This paper presents a model concept that uses the clay shrinkage characteristics to derive crack volume and crack depth under transient ®eld conditions. The concept has been developed to simulate ®eld average behaviour of a ®eld with cracks, rather than ¯ow and transport at a small plot. Water ¯ow and solute transport are described with basic physics, which allow process and scenario analysis. The model concept is part of the more general agrohydrological model SWAP, and is applied to a ®eld experiment on a cracked clay soil, at which water ¯ow and bromide transport were measured during 572 days. A single domain model was not able to mimic the ®eld-average water ¯ow and solute transport. Incorporation of the crack concept considerably improved the simulation of water content and bromide leaching to the groundwater. Still deviations existed between the measured and simulated bromide concentration pro®les. The model did not reproduce the observed bromide retardation in the top layer and the high bromide dispersion resulting from water in®ltration at various soil depths. A sensitivity analysis showed that the amounts of bromide leached were especially sensitive to the saturated hydraulic conductivity of the top layer, the solute transfer from the soil matrix to crack water ¯ow and the mean residence time of rapid drainage. The shrinkage characteristic and the soil hydraulic properties of the clay matrix showed a low sensitivity.