In sub Saharan Africa agricultural production is low and considered a main focus for development as concerns about food security and crop yield are evident and currently became even more emphasized by the impact of climate change. Over the years agronomic research definitely has been successful in identifying crop productivity constraints and addressing these by technological solutions and recommended practices. Still, real impact on crop yield appears limited, despite numerous efforts to support effective implementation of such recommendations. Farmers in many cases hesitated to adopt the technologies proposed, for example, as they considered them too risky. At the same time, differences in backgrounds of farmers and researchers resulted in different choices made and consequently in different pathways to development. Participatory experimentation, in which farmers and researchers co-operate, is often proposed as an interesting option to increase understanding between researchers and scientists and considered well suited to develop options to upgrade farming systems within a local context. Involving farmers in the development of recommendations matching with local conditions is assumed to make future implementation more likely and sustainable. Furthermore, farmers’ involvement in participatory experimentation is expected to result in empowerment. The impact of participatory experimentation in the context of increasing crop productivity can be divided into two components: (1) contributions to higher yield and sustainability of the agricultural production system and (2) contributions to the social and human capital of the farmers involved. In our research project in Tigray, Northern Ethiopia, effectiveness of participatory experimentation was assessed with respect to these contributions and contextual relevance. Furthermore, better understanding of the participatory process itself and the agronomic outcomes of the experiments was sought. In our research project 16 groups of farmers, distributed over four locations, were involved for four years in participatory experimentation. Next to monitoring the experiments through various measurements, also the process and the farmers involved were monitored through interviews and observation. Participatory approaches differ in the level of control of the process by the farmers involved; in our research project an approach was followed in which farmers were unambiguously in the lead: they selected the technologies they wanted to assess within their local context; researchers involved concentrated on facilitation of the participatory process. Identification of constraints and opportunities is an essential step in most research processes, at the same time methods applied are diverse. In a review of 16 papers we identified three clusters of methods on the basis of two variables typifying these methods: control over de research process and represented opinion. In a case study we compared, as a follow up, three different methods for identifying crop productivity constraints: contextual data collection, individual surveys and focus group discussion. Congruency between the methods was not found significant, and of these methods only focus group discussion responded to contextual diversity. Combined the review and case study demonstrated that process control and represented opinion had a manifest impact on generated outcomes. In our study area we applied focus group discussion in all four locations to identify productivity constraints and opportunities. Outcomes in the form of mind maps were quantified to allow comparison between the locations. In all locations soil fertility measures were considered a main opportunity; all other categories of constraints and opportunities were diverse for the locations involved. Farmers were in groups involved in participatory on-farm experimentation to arrive at best practices matching with local preferences, complexity and context. Experiments conducted were documented for yield and various soil, land and management related factors. Analysing the outcomes of the on-farm experiments indicated that the (high) variability in grain yield observed was very location specific and related to treatment effects, local management, climate and soil conditions. Furthermore, relationships between different soil properties (organic-C, P and K) and response to fertilizer inputs were indeterminate, but this was not the case for N-total. Developing clear recommendations on the basis of the experiments consequently was not possible. Understanding choices made by farmers in experimentation processes is important to find reasons why farmers often hesitate to adopt specific practices. We therefore monitored the 16 groups (of five farmers) during four years and found that the groups followed a very rational context-rooted strategy that differed considerably from that of the researchers involved. Consequently, in participatory experimentation, involvement of farmers in defining the actual experimental design is mandatory in order to deal with local preferences and context. Experimental outcomes in the form of yield responses and nutrient balances were analysed from different perspectives: farmer, agronomic and environmental. All three perspectives indicated that gradually strengthening the existing farming system by using fertilizers, organic manure and legume fallows will support crop productivity while addressing at the same time other aspects of sustainability like food security and profitability. Farmers were involved in participatory experimentation for four years with minimum external intervention; in contrast to our initial expectations, all groups continued their involvement and indicated the ambition to proceed on their own. Of a set of factors that might influence farmers’ involvement, only benefits in the form of good responses were overall important; all other factors were highly variable among the groups. Outcomes achieved for the farmers involved were found substantial and relevant; yield increase, knowledge and confidence being most important. Next to this, participatory experimentation resulted in better mutual understanding of perspectives held by farmers and scientists, reducing in this way the gap between their views. To arrive at effective participation, approaches should be group specific and based on open processes in which feedback is essential and responsibilities are delegated to the farmers in all research phases. Facilitation is essential in this but given the diversity of groups and the context in which they operate, blue-print approaches are not likely to be effective. Human-social outcomes and contextual relevancy were extremely influenced by an explicit choice for maximum involvement of farmers in all phases of the experimentation process; farmers’ learning consequently not only related to knowing the best way, but also towards confidence in finding the best way. Participatory experimentation, therefore, not only constitutes a feasible option to improve farmer livelihoods by combining development of site specific recommendations with capacity building, but also has a clear potential to act as a change agent to trigger transitions towards more sustainable livelihood systems.