Four 10‐ha plots in dense watermilfoil beds of Lake Seminole, Georgia, were each treated with either 2,4‐D DMA or 2,4‐D BEE at rates of 22.5 and 45 kg a.e./ha. Both formulations were shown to be rapidly converted to the 2,4‐D acid form, with no detection of 2,4‐D DMA or 2,4‐D BEE in the water within less than 24 hours after treatment. The maximum detected 2,4‐D concentrations in the high rate 2,4‐D DMA and 2,4‐D BEE plots were 3.6 and 0.68 mg/, respectively. However, all but seven samples at a 2,4‐D BEE plot showed nondetectable herbicide levels by day 7, with all water samples showing nondetectable levels by day 13. Dimethylnitrosamine and 2,4‐dichlorophenol, potentially toxic transformation products of the herbicide formulations, were at nondetectable levels in all water samples. Sediment samples showed no significant net accumulation of 2,4‐D, 2,4‐D BEE, or 2,4‐dichlorophenol during the summer monitoring; dimethylnitrosamine remained at nondetectable levels. There was no accumulation of 2,4‐D in fish collected from the two plots treated with 2,4‐D DMA. Four of 24 game fish from the 2,4‐D BEE treatment plots contained low levels of 2,4‐D in muscle tissue, with a maximum value of 0.29 μg/g. In contrast, 18 of 20 gizzard shad collected from these plots through day 13 contained detectable 2,4‐D in the muscle, with a maximum concentration of 6.9 μg/g. All fish collected after day 13 contained nondetectable levels of 2,4‐D. Small decreases in dissolved oxygen and pH, associated with the complete watermilfoil control in all plots, had returned to normal summer values by day 28.