Plots 17 × 5 m were laid out in farmers' fields in three villages. Treatments (replicated thrice) included atrazine (0.5 or 1.0 kg a.i./ha) on sorghum and nitrofen (0.75 and 1.5 kg a.i./ha) on groundnuts with or without hand-weeding or farmers normal weed control. Groundnuts were unfertilized, sorghum received 20 or 80 kg N/ha. Cooperating farmers were allocated plots for traditional weed control, other operations were carried out by research staff. Some plots were discounted because farmers withdrew cooperation or did not participate correctly or because of flooding. In groundnuts nitrofen alone was inferior to nitrofen + one hand-weeding. In sorghum there were no differences between high and low fertilizer regimes. Atrazine gave superior yields to farmers' weed control. All treatments were better than the unweeded control. Generally farmers fields had low weed populations due to good, timely control and low soil fertility. Herbicides are not economic at present. Problems associated with research in farmers fields included: selection of farmers who would participate correctly, variability between different farmers' weeding practices and between different fields due to previous crops and cultural practices; lack of control of farmers weeding operations and the distance of experimental plots from the research station. The authors suggest that small experiments with only 4-6 treatments are easier for participating farmers to comprehend, selecting suitable size and shape of plots is important and that if many farmers are involved the number of replicates can be reduced.