In two trials cassava (Manihot esculenta Crantz; 'Karangwa', a bitter variety) planted at various densities under rainfed conditions was intercropped with groundnut (Arachis hypogaea L.; 'Makulu Red') and upland rice (Oryza sativa L.; local variety). For the groundnut-cassava trial, the cassava was harvested after 10 months and 16 months. It was demonstrated that cassava at densities as high as 1 7 000 plants ha-1 had no significant effect on groundnut yields and that there was little difference between cassava yield at that density and at 5500 plants ha-1. Cassava alone yielded about twice that of intercropped cassava, but groundnut yields were similar when monocropped and intercropped. Leaving cassava in the ground for 16 months, rather than 10, resulted in trebling of cassava yield. Cassava growth and yield were higher when intercropped with groundnut than with rice, but rice yielded similarly when monocropped and intercropped with cassava planted at various densities.