Movement and persistence of simazine and atrazine in an acid sand were measured in 1989 at a field site that received 76-105 mm of rainfall during 100 days after herbicide application. Three environments were created: (i) warm and wet (early herbicide application of 100 mg m(-2) on 28 May); (ii) cool and dry (later herbicide application of 100 mg m(-2) on 24 June); (iii) cool, dry and high pH (the later herbicide application after the surface pH had been raised from 4 . 8 to 6 . 4 by the application of 150 g m(-2) CaCO3 42 days earlier). The soil profile was sampled for herbicide residues and soil water at six depths on five occasions during 100 days after herbicide application. Rainfall, pan evaporation, and soil and air temperatures were measured throughout the experiment. Soil temperatures at 13 mm ranged from 11 to 41 degrees C (daily minimum and maximum). Most of the herbicide remained in the surface 25 mm of soil and after 100 days about 25% of that applied remained in the soil profile in the acid soils and about 40% remained in the soils treated with CaCO3. Despite eight, or nine, leaching rains that a simulation model estimated would have wetted the profile to 300 mm, residues <2 mg m(-2) were found in the 150-200 mm layer and in the high pH treatment only. A simulation model gave good agreement with observed losses of simazine residues in soil maintained at field capacity and buried within the soil profile. However, the simulation model underestimated residue losses in the soil profile and indicated that processes other than chemical hydrolysis were involved in the degradation of herbicide in the surface 25 mm of soil. Weed control reliant on persistent residues would be prolonged if the herbicide is applied near the sowing date of the lupin crop and incorporated below the surface 25 mm of soil by the method of sowing.