Field experiments were conducted at the Irrigation Research Station, Kadawa in the Sudan savanna zone of Nigeria during the 1984 and 1985 dry seasons to identify selective pre‐emergence herbicide mixtures for weed control in two varieties of okra(Abelmoschus esculentusL. Moench). The variety of okra only had a significant effect on crop vigour score at 6 weeks after sowing in 1985 when var. V35 was more vigorous than TAE‐38. In 1984, despite supplementary hoe‐weeding at 6 weeks after sowing, all the herbicide treatments resulted in significantly lower fruit yield than the hoe‐weeded control. Among the herbicide treatments, metolachlor plus prometryne at 2.0 + 1.0 kg a.i./ha followed by supplementary hoe‐weeding resulted in significant weed reduction and the highest okra fruit yield. Metolachlor plus terbutryne and pendimethalin plus prometryne, each at 2.0 + 1.0 kg a.i./ha. followed by supplementary hoe‐weeding also reduced weed infestation significantly and resulted in fruit yield comparable to that of the best herbicide treatment. In 1985 only metolachlor plus prometryne at 2.0 + 1·lb0 kg a.i./ha followed by supplementary hoe‐weeding gave comparable okra fruit yield to the hoe‐weeded control. Compared with pre‐emergence application of herbicides, directed post‐emergence herbicide treatments resulted in significant crop growth depression and unacceptably low okra fruit yields. Losses in fruit yields resulting from uncontrolled weed growth were 99.1% and 100% compared with the hoe‐weeded controls in 1984 and 1985, respectively.