Velvet bean [Mucuna deeringiana (Bort.) Meer.], pigeon pea [Cajanus cajan (L.) Huth], sorghum [Sorghum bicolor (L.) Moench.] and tropical pumpkin or calabaza [Cucurbita moschata (Duchesne) Poir.] were evaluated as cover crops for the control of nutsedges in rotation with onion (Allium cepa L.) at the Lajas Agricultural Experiment Station during the years 1998-99 and 1999-2000. In each year of study, the four cover crops were grown until mature stage, and plant residue was disked or removed from soil surface before onion planting. Cover crop species had no significant influence (P < 0.05) on nutsedge density, mainly represented by Cyperus rotundus L. and C. esculentus L., neither six weeks before onion planting nor after nine weeks of cropping during 1998-99. Disc incorporation of all cover crops suppressed nutsedge density more than removal of plant residues from soil surface. Onion produced greater yield (30,030 kg/ha) after calabaza rotation than after pigeon pea (21,090 kg/ha) or sorghum (18,940 kg/ha) in 1998-99. In 1999-2000, plots grown with velvet bean, pigeon pea and calabaza had less nutsedge than the untreated controls two weeks before incorporation of plant residues. Plots with these three cover crops also had lower density of nutsedges than plots with sorghum. Nutsedge density was not significant at three, six, and 10 weeks after onion planting. In 1999-2000, onion yields among cover crop rotations were not significantly different, with an average of 5,837 kg/ha. Cover crop allelopathy, as well as nutsedge interference, may have reduced onion production during the second year of planting.