Farmer-participatory research (FPR) can help improve the effectiveness of technology development, raise adoption rates, and increase the payoff to agricultural research. However, there is wide diversity of opinion about the scope and nature of farmer participation in the implementation of FPR. This study compared different FPR approaches being used by various organizations at six on-farm sites in Malawi and Zimbabwe, with a view to developing guidelines for future work, particularly in developing technologies targeted at women farmers. The study found differences between the various approaches in diagnostic activities, planning of experiments, assessment of results, and strengthening of farmers' capacity to conduct their own experiments. Overall, however, the various methods were not completely distinct, but represented a continuum from traditional research-led, to researcher-led with farmer input, to farmer-led with research input. The results indicated that when there are no clear procedures to directly target women farmers they tend to be under-represented, and gender issues are not sufficiently integrated into the research process. PRA techniques were also found to be useful in planning experiments, but require considerable investment of time by researchers and extension staff. Researchers therefore need to formulate clear objectives for conducting different types of on-farm experiments, ensure that farmers understand their roles in each case, and finally integrate different types of research to develop technologies that are practical and profitable.