The phytotoxicities of soil-incorporated 2,6-dichlorobenzonitrile [dichlobenil] and 2,6-dichlorothiobenzamide [hereinafter referred to as SD-7961] were compared in five soils in the greenhouse. Dichlobenil and SD-7961 were about equally toxic to snap beans (Phaseolus vulgarus L., var. Black Valentine) in each of the five soils. The herbicides were more toxic in Christiana loam than in the other four soils. Phytotoxic residues of both herbicides disappeared most rapidly from Hagerstown silty clay loam and Christiana loam and least rapidly from Sharkey clay. Gas chromatographic analyses of soil extracts confirmed that dichlobenil was most persistent in the clay. Small amounts (0.14 ppmw or less) of 2,6-dichlorobenzoic acid [hereinafter referred to as 2,6-DBA] were detected in soil from greenhouse pot cultures about 9 months after incorporation of dichlobenil. In a field experiment at Beltsville, Maryland, residues of dichlobenil persisted from one season to the next but did not build up in soil sprayed annually for 3 years with 4 or 8 lb/A. Some movement from the surface 6 in into the 6 to 12-in zone occurred on plots sprayed 2 years in succession with 8 lb/A or once with 40 lb/A. Rapid loss of dichlobenil occurred during summer months. When the concentration of dichlobenil on the 40-lb/A plots dropped to about 10%, subsequent loss occurred at a very slow rate.