It was studied the feasibility of reducing there commended herbicide rate, single or in mixtures, without affecting the yield or other desirable plant features, for cultivar Santa Rosa in "Dark-Red Latossol -sandy phase". ' The trials was settled at randomized block design with treatments repplicated 3 times, testing the full recommended rate, 25% an d 50% reduction of full rate of trifluralin, alachlor and metribuzin alone and in mixtures. There - commended rates were 8, 06 ; 1, 72 and 0,28 kg/h a of trifluralin, alachlor and metribuzin, respectivelly. Mixtures with reduced rates of trigluralin metribuzin (0,6 5 + 0,21 kg/ha) and alachlor + metribuzin (1,44 - 0,21 kg/ha ) presented control rate higher than 90% up to 60th day after sowing without any phytotoxicity and deleterious effects on no dulation. Weed control achieved by 25% r eduction in herbicide rates was equivalent to hoeing and afforded similar levels of protein , et her -extract and ash content, weigh of 100 grains, number of grains per pod, plant heigh and height of inferior pod insertion. The highest soybean yield, achieved wi th full rates of the mixtures were statistically similar to yields provided with 25% reduction of the recommended rates of same herbicides. Therefore, in accordance to economic and ecologic conveniencies, it is suggested that the actual rate should be reduced 25%, under the conditions of this trial.
Soybean; herbicides; Low - rates; yield