Bioassays with soybeans and lentils were used to compare the persistence of 2,4,5-T, triclopyr, and picloram in soil after applying 0.038, 0.38, and 3.8 kg ae/ha under field conditions. Soil samples were collected from 1 to 269 days after spraying and were kept frozen until growth room bioassays were conducted. Triclopyr was slightly less persistent than 2,4,5-T, but neither herbicide was biologically active during the next season. At least 90% of picloram disappeared within 7 months, but its biological activity was more persistent than that of either triclopyr or 2,4,5-T. Nine months after treatment, neither lentils nor soybeans could be grown in soil treated with picloram at 3.8 kg/ka. Triclopyr and 2,4-D had smiilar soil thin layer chromatographic mobilities in four different scils. Picloram was significantly more mobile than either triclopyr or 2,4,5-T, primarily due to its lower adsorption in the soils examined.