-The aim of this study was to assess the efficacy of S-metolachlor applied in pre-emergence conditions for the control of Brachiaria decumbens, Digitaria horizontalis, and Panicum maximum in sugar cane mechanically harvested without previous burning of the crop (green harvest) with the crop residue either left or not on the soil surface. The experiments were established in the field according to a randomized complete block design with four repetitions in a 7 x 2 split-plot scheme. In the plots, five herbicide treatments were studied (S-metolachlor at 1.44, 1.92, and 2.40 kg ha , and isoxaflutole at 0.188 kg ha -1 ), and two control treatments with no herbicide application. In the subplots, the presence or absence of sugar cane crop residue on the soil surface was evaluated. S-metolachlor efficacy was not hampered by either 14 or 20 t ha -1 of sugar cane crop residue on the soil surface. When sugar cane crop residue was covering the soil surface, S-metolachlor at a rate of 1.44 kg ha -1 resulted in weed control similar at their larger rates, where as without the presence of crop residue, S-metolachlor controlled B. decumbens, D. horizontalis, and P. maximum at the rates of 1.92, 1.44, and 1.92 kg ha -1 , respectively. The herbicides clomazone and isoxaflutole were effective for the studied species, independently of the crop residue covering the soil surface. S-metolachlor caused no visible injury symptoms to the sugar cane plant. Clomazone and isoxaflutole caused visible injuries to the sugar cane plant. None of the herbicides negatively affected the number of viable culms m 2 or the culm height and diameter.Keywords: signal grass, crabgrass, guinea grass, crop residue.
RESUMO -Objetivou-se com este trabalho avaliar o controle em pré-emergência de