Herbicides were evaluated for selective weed control in pearl millet in field trials conducted in the Sudan and the Northern Guinea zones of Nigeria from 1977 to 1979. Some of the herbicide treatments had a supplementary weeding, during the two later years of the study. The work was extended to the Southern Guinea zone in 1979. Propazine at 0.8 kg a.i./ha was the most tolerated herbicide on the loamy sand soils of the Sudan zone. It was tolerated up to 1.6 kg a.i./ha in the comparatively heavier sandy loam soils of the Guinea zones. Linuron was toxic to the crop at equivalent rates. Mixtures (1:1) of the two herbicides were better tolerated than linuron alone but did not show any advantage over propazine. The selectivity of terbuthylazine, used in one out of three years, was comparable in the Northern and Southern Guinea zones while it was less selective in the Sudan zone. Under severe weed infestation, one supplementary weeding following pre-emergence herbicides resulted in yields comparable to those obtained with two hoe-weedings. An average loss of 46% in grain yield was caused by unchecked weed growth.