ABSTRACT In view of the adverse impact of weeds on crop yields, the resource-limited subsistence farmers in the northern savanna zone of Ghana appreciate the benefits of maintaining weed-free farms. A field experiment to assess the types and levels of weed infestation in maize via a range of tillage systems including hand hoe, bullock, tractor, and zero tillage as control was conducted at Nyankpala in 1996 and 1997. The experiment was laid in a randomized complete block design with four replications. Compound fertilizer (20-20-20) was applied at planting giving 60 kg N ha−1, 60 kg P ha−1 and 40 kg K ha−1. This was top dressed with sulphate of ammonia representing another 60 kg N kg ha−1. Weed control was by hand hoeing at three and eight weeks after planting of maize. The results revealed that the pre-planting application of glyphosate (N-phosphonomethyl glycine) on zero-tillage significantly (p < 0.05) suppressed weed infestation at three weeks after planting (3 WAP) by an average of 97.3%. However, at harvest, in the absence of weed control in crops, the weed populations and weed dry matter on zero-tillage were similar to the hand hoe treatment. Grass weed populations were not affected by the tillage systems. Broad-leaved weed populations on the hand hoe and zero-tillage systems were similar but significantly higher (p < 0.05) than on the tractor and bullock treatments. The hand hoe and the zero-tillage maintained a balance between the broad-leaved and the grass weeds but the application of the tractor and bullock treatments produced statistically higher (p < 0.05) grass than broad-leaved weeds. Zero-tillage significantly (p < 0.05) reduced maize plant height and grain yield by 4% and 30%, respectively, compared with other tillage systems, which had similar heights and grain yields. Thus, the bullock tillage system can be considered the most appropriate choice for peasant farmers in northern Ghana, given the associated cost of herbicides, tractor services, and the labour requirements of hand hoeing.