We evaluated four chemical treatments (herbicides) used to fight weeds, as compared to manual weed picking and using a witness treatment. All treatments had a rotational handling with 6.5 animal units (12 animals of 250 kg per five haurs). We used a proteid bank or monoculture of forage peanut, in an arca of 1080 sq.mt., with a 30 days break. In order to evaluate the many variables, we used an experimen-tal design of complete random blocks, with six treatments and tree repetitions per treatment. In order to determine the coverage of the culture, its tolerance to the active ingredient and the level of treatment control, we carried out four evaluations: 0, 30, 60, and 90 days of treatment application. We found differences (P<0.01) between treatments related to the degree of coverage of the forage peanut plants. With Gliphosate and manual weeding, we obtained a dominant coverage of forage peanut, while with the witness treatment the degree of coverage was inferior. The production of dry matter from the weeds, as well as their percentage in the pasture, decreased in a significant way (P<0.01), through the sampling process. With the witness treatment we obtained the largest amount of dry matter in the weeds, and also the highest percentage of weeds in the pasture.