Participatory approaches are advocated as being more effective in supporting rural development processes than traditional top-down extension approaches. Participatory experimentation involving both farmers and researchers is often expected to result in processes of experiential learning. Assuming that such learning leads to change in farmers' views and practices, we wanted to identify these changes. For that purpose we applied an analytical framework that included three dimensions (process, outcomes, impact) and functional as well as human–social aspects. We involved farmers in group-based participatory experimentation for four years with minimum external intervention, aiming for maximum control of the experiments by the farmers themselves. In total 16 groups of farmers divided over four locations participated. Data were derived from interviews and observations. In general participants considered their participation worthwhile and mostly valued learning-aspects. Farmers indicated that they acquired new knowledge and became confident with respect to specific agricultural practices such as fertilizer application. They also felt more confident in conducting systematic experimentation. This confidence is supported by our observation that they managed to achieve positive yield responses, over 50% in most cases. Participating farmers responded significantly differently after the four years of experimentation compared to a control group of local farmers. After the four years they would: (1) involve non-family more in their discussions about farm management; (2) address officials more easily to solve neighbourhood problems; and (3) be more specific in their ambitions to learn about agriculture. Participants perceived significantly more (positive) change towards productivity and poverty reduction compared to the control group. In contrast to our initial expectations, all groups continued their involvement in the experiments for four years and indicated the ambition to continue on their own. Of a set of factors that might influence involvement of farmers, only benefits in the form of good responses were overall important. All other factors were highly variable among the groups. We concluded that change was achieved with respect to functional and human–social aspects, which are both essential components of agricultural systems and affect their transformation. In designing processes of participatory experimentation it is, therefore, important to take such non-uniform sets of impact factors into careful consideration. Given the diversity of groups and the context in which they operate, blue-print approaches are not likely to be effective due to insufficient incorporation of local group variability.