Leaching of atrazine [6-chloro-N-ethyl-N '-(methylethyl)-1 ,3,5-triazine-2,4-diamine] and terbuthylazine [6-chloro-N-(1, 1-dimethylethyl)-N'-ethyl-1 ,3,5-triazine-2,4-diamine] was studied in 23 soils by means of the soil thick-layer leaching method. The equivalent of 4 kg ai ha-1 0f both herbicides was leached with water equivalent to 129 mm rain. Distances leached were measured qualitatively in bioassays with the test plant Trifolium repens L. Leaching distances in different soils varied from 6-20 cm for atrazine, and from 2-16 cm for terbuthylazine. Atrazine leached at least 10 cm in 13 soils, while terbuthylazine did not move from the zone of application in 12 soils . Of the five soil properties assessed by means of simple regression, organic matter content (% C) and P-reversion were significantly correlated with the leaching distance of atrazine, and % C with the mobility of terbuthylazine . Relationships between herbicide leaching and either total clay content or soil pH were relatively weak. The combination of organic matter content and soil pH in multiple regression analysis gave the best prediction of leaching for atrazine (R = 0.85) and terbuthylazine (R = 0.42). It is suggested that the regression models incorporating % C and pH be used for rapid ranking of soils regarding their leaching indices for the two compounds . The models may prove useful for preliminary identification of soils which are prone to leaching of the herbicides.Die loging van atrasien [6-chloor-N-etiel-N' -(metieletiel)-1 ,3,5-triasien-2,4-diamien) en terbutielasien [6-chloor-N-(1, 1-dimetieletiel)-N'-etiel-1 ,3,5-triasien-2,4-diamien) is in 23 grondsoorte deur mid del van die grond-diklaagmetode ondersoek. Die ekwivalent van 4 kg ab ha-1 van beide onkruiddoders is met 'n volume water ekwivalent aan 129 mm reen geloog . Logingsafstande in die verskillende gronde is in biotoetse met die toetsplant Trifolium rep ens L. gemeet. Logingsafstande in verskillende gronde het vanaf 6-20 cm vir atrasien, en 2-16 cm vir terbutielasien , gevarieer. Eersgenoemde onkruiddoder het minstens 10 cm in 13 gronde geloog, terwyl laasgenoemde by 12 gronde nie uit die sone van toediening beweeg het nie. Van die vyf grondeienskappe wat deur middel van eenvoudige regressie beoordeel is, was organiese materiaalinhoud (% C) en P-reversie betekenisvol met die logingsafstand van atrasien, en % C met die loging van terbutielasien , gekorreleer. Vir beide onkruiddoders was die verwantskappe tussen loging en totale klei-inhoud 6f grond-pH relatief swak. Die kombinasie van organiese materiaalinhoud en grond-pH in 'n meervoudige regressievergelyking het die loging van atrasien (R = 0.85) en terbutielasien (R = 0.42) die beste voorspel. Hierdie modelle word vir voorlopige indeling van gronde volgens hul logingsindekse vir atrasien en terbutielasien aanbeveel. Die mode lie mag nuttig wees vir identifisering van gronde waar die logingspotensiaal van die twee onkruiddoders hoog is.