On-farm trials to determine the optimum combination of maize ( Zea mays
L.) density, bean ( Paseolus vulgaris
L.) cultivar and bean spatial arrangement to produce high yields of the intercrop combination were conducted in Chinyika Resettlement Area (CRA) and at Domboshava Training Centre (DTC) during the 1996/97 and 1997/98 rainy seasons. The effects of maize at 37, 000 and 24, 000 plants ha-1
, bean cultivars 'Natal Sugar' and 'Carioca', and bean arranged in one or two rows between rows of maize or bean planted in the same row as maize were evaluated in a completed factorial arrangement. Maize density, bean cultivar, bean spatial arrangement and their interactions significantly (P<0.01) affected maize and bean grain yields at both locations for the two seasons. Changing the maize density from 90 x 45 cm (24000 plants ha-1
) to 90 x 30 cm (37000 plants ha-1
) increased maize yield by 28 and 39% and reduced bean yields by 11 and 18% in the respective seasons. Maize yield was 19% less when intercropped with Natal Sugar than with Carioca. Same row and one row bean cultivar arrangements produced similar maize yields. Land equivalent ratios (LER) were greater than one for all but one intercrop arrangements at CRA but less frequently at DTC. The maize density of 90 x 30 cm with Carioca in the same row as maize produced the highest LER value at all sites, 1.73 at Chinyudze, and 1.53 at Domboshava. The same treatment gave the highest gross return ofZ$12 649 ha-1
at CRA. Carioca planted in the same row as maize at the maize density of 37,000 plants ha-1
is the most ideal approach to dryland maize/bean intercropping since it achieved high yields and allows easy weeding.