The objective of this meta-analysis of evaluations is to ascertain factors and interventions that do or do not lead to gender-related social norm change, identify the reasons behind failures, and provide knowledge about gender programming in the agricultural sector. A meta-analysis was conducted on 41 documents, including 18 evaluation reports. Following a rigorous two-stage process only three evaluation reports reached the deep dive stage. The analysis suggests that many evaluations lack methodological rigor, and fail to capture gender norm changes. Moreover, many projects do not apply the lessons they learn and don’t share these lessons with partners or other stakeholders. The findings identify factors that contribute to changes in social norms, including a multi-intervention design, relevance of the interventions to community needs, use of social mobilization, and initiatives to overcome constraints to women’s participation. Social accountability tools that are designed to improve participation and equality result in greater participation of women in project activities. Gender programming can be improved in the following ways: ensuring gender related indicators are set and that the theory of change includes gender norm transformation; using participatory planning and social accountability methods; working with women in groups; adopting a multi-intervention design that aims to affect change at different levels sequentially; taking a long term approach to change and project design; training project teams in gender awareness; documenting gender learning and disseminating this knowledge widely; and, incorporating such knowledge into new project designs.