The only commercial genetically modified (GM) subsistence food crop is white maize in South Africa, which was released in 2001/2. This paper reports on the performance of insect resistant (Bt) white maize grown by smallholders in Hlabisa, KwaZulu Natal, where the other development is minimum tillage. The results show that, contrary to many inflated claims, in the dry 2003/4 season, there was no significant difference between the yield per kg of seed for Bt and conventional maize, due to very low stalk borer infestation levels. Farmers who planted Bt maize in 2003/2004 were thus worse off as they paid more for seed and obtained no benefit. This is measured using efficiency scores from a stochastic frontier analysis. These results conflict with the yield per hectare figures, which show a gain of 25% from using Bt. We think this shows that kgs per hectare is not a meaningful performance measure for African smallholders. More interesting is the effect of minimum tillage, which apart from reducing erosion, increased yield per ha by 12%, while reducing costs and increasing efficiency by 9%. But the saving is in reduced labour, which may not be an advantage if jobs are lost.