Two field trials with garlic 'Lavínia' and onion 'Monte Alegre' and 'Roxa do Traviú' were performed on a loamy soil to study herbicide selectivity and their efficiency on weed control. The following treatments were studied on garlic: pendimethalin at 1.50 kg/ha; linuron at 1.00 kg/ha and oxadiazon at 1.00 kg/ha, applied in pre-emergence at the planting day; ioxynil at 0.50 and 0.75 kg/ha, ioxynil + oxadiazon at 0.75 + 0,25 kg/ha and 0.50 + 0.50 kg/ha, applied 35 days after, weeded and hoed checks and a check with mulch. The weed population was a mixture of various types of weeds; predominated by Portucala oleracea L. Weed control was evaluated at 48 and 84 days. All treatments controlled very well the dicotyledons. Ioxynil alone or mixed with oxadiazon in post-emergence did not control grasses. None of the herbicide treatments reduced the garlic stand, occurring the lowest number on the checks with weeds and mulch. The yields occurred in the following: hoed check > oxadiazon > ioxynil + oxadiazon at 0.50 + 0.50 kg/ha = pendimethalin = mulch = ioxynil + oxadiazon at 0.75 + 0.25 kg/ha > linuron > ioxynil at 0.75 kg/ha > ioxynil at 0.50 kg/ha > weeded check. On the onion crop the following treatments were studied: pendimethalin at 1.50 kg/ha, linuron at 1.00 kg/ha, oxadiazon at 1.00 kg/ha, and chloroxuron at 4.00 kg/ha, applied in pre-emergence on the day of transplanting; ioxynil at 0.50 and 0.75 kg/ha, oxadiazon + ioxynil at 0.75 + 0.25 kg/ha, at 1.00 + 0.25 kg/ha and at 1.00 + 1.00 kg/ha and linuron at 1.00 kg/ha, all applied 32 days after, weeded and hoed checks. The weeds occurring were predominantly dicotyledons being Portulaca oleracea L. and Chenopodium album L. the most dominant. Control evaluation was done at 48 days showing an excelent control of dicotyledons (over 95%) by all treatments. Ioxynil alone or mixed with oxadiazon did not control grasses and chloroxuron showed the lowest general control but yielded satisfactory. Oxadiazon in pre-emergence and linuron in post-emergence were the most efficient in controling gresses. There were no differences among treatments for bulb yield.