This paper presents an application of a two-domain model of water flow and solute transport in macroporous soil (MACRO) to field experiments in drained and irrigated, saline heavy clay soils under cotton. Model predictions are compared to detailed measurements of the soil water balance and leaching of chloride to field drains made during a 17-day period following two successive sprinkler irrigations. The model was able to reproduce the measured drain hydrograph and the observed response of the water table, providing calibrated, rather than directly measured, saturated hydraulic conductivity values were used. At the end of the first irrigation, the soil profile was fully recharged, with the water table only 10 cm below the surface. In the 2 weeks following irrigation, soil water extraction by the crop was largely restricted to the upper 30 cm of soil, presumably due to excessive salt concentrations in the subsoil. Indeed, the model simulations indicated a reduction in transpiration below the potential rate after only 1 week, with an accumulated water deficit of 60 to 70 mm. A strong dilution of the chloride concentrations during peak drain discharges was observed and was also predicted by the model. This dilution was related to the rapid infiltration of irrigation water of low salt concentration in the cracks. In two flow domains, the model precisely matched observed leaching losses of chloride during peak drain discharges, but underestimated by ≈ 25% the accumulated loss of chloride 1 week after the irrigation. This was entirely due to an underestimation of chloride concentrations in the outflow during the late stages of recession. During such recession flows, the chloride concentrations in the drain outflow (≈ g· l −1) were most likely influenced by inflows of saline shallow groundwater from surrounding areas. In one flow domain, the model predicted qualitatively similar patterns of drain outflow and chloride leaching. Nevertheless, the model performed less well when bypass flow was not taken into account, with chloride leaching overestimated during peak discharges, despite a total drain outflow which was 10 mm less than in the two-domain case.