Four forest watersheds (1 ha each) in the upper piedmont of Georgia were treated with hexazinone [3-cyclohexyl-6-(dimethylamino)-1-methyl-1,3,5-triazine-2,4(1H, 3H)-dione] pellets3 at a rate of 1.68 kg ai/ha. From the end of April, 1979, until May, 1980, 26 storms were monitored to determine movement of hexazinone and two of its metabolites [A: 3-(4-hydroxycyclohexyl)-6-(dimethylamino)-1-methyl-1,3,5-triazine-2,4(1H, 3H)-dione, and B: 3-cyclohexyl-6-(methylamino)-1-methyl-1,3,5-triazine-2,4(1H, 3H)-dione] in runoff water. Residues in runoff peaked in the first storm after application (mean concentration of 442 ± 53 ppbw), and declined with subsequent storms in a power curve function: Conc. (ppbw) = 405 × rate × (1 + 0.44 × days)-1.1. Loss of hexazinone in storm runoff averaged 0.53% of the applied herbicide, with Storms 1 and 17 accounting for 59.3% of the chemical exported. Storm 1 had high residue concentrations and low runoff volume, while Storm 17 contained only low residue levels but a very large stormflow. Hexazinone was the predominant compound in the runoff of all 26 storms. Metabolites A and B occurred in runoff in low-to-trace concentrations (<23 ppbw) for up to 7 months after application. Subsurface movement of hexazinone appeared in streamflow 3 to 4 months after application and produced an additional loss of 0.05%. A second-order perennial stream below the treated watersheds periodically contained hexazinone residues of <44 ppbw.