Summary Smallholders face numerous constraints that keep them from taking advantage of market opportunities. Because they often live in poverty and remoteness, on less than a hectare of land with poor road and market infrastructure, they face high costs of market exchange. In order for smallholders to compete with large-scale farms and benefit from the observed and ongoing transformation of the agri-food sector, institutional solutions are necessary that address the small scale problem. One potential option is through farmer collective action. However, the existing literature has not yet sufficiently identified the explicit mechanisms and dynamics by which collective action generates benefits for men and women. In three related articles, this research investigates how farmer groups facilitate access to input and output markets by reducing transaction costs, thereby promoting intensification and commercialization of small farms. It addresses the existing research gap by analyzing determinants, participation dynamics and impacts of farmer groups, using the case of small-scale banana producers in Kenya. For the analysis cross-sectional data from 444 farm households was collected in the central highlands region of Kenya, where bananas provide an important source of food and income. Recently established farmer groups initiated by two local nongovernmental organizations (NGOs), TechnoServe and Africa Harvest, introduced institutional change in the national banana market chain. Within such groups, farmers can more easily access clean tissue culture (TC) banana planting material, get agronomic and market information, negotiate better prices and gain access to urban, high-value markets through collective selling. However, the marketing performance of these groups is unclear, and commitment varies greatly between members. Gender issues are partly left unaddressed by NGOs and group leadership. These issues pose serious threats to the groups’ viability in the long run. The first chapter focuses on the determinants and impacts of smallholder organization. Farmer groups are generally inclusive of the poor, although wealthier households are more likely to join. Access to credit plays a role, as the adoption of relatively costly TC planting material is technically encouraged among group members. Distance to paved roads and mobile phone ownership facilitate participation and thus have a positive effect on the likelihood of group membership. Employing a propensity score matching method to reduce possible selection bias in impact assessment, we show positive income effects for group members. However, the magnitude of the economic benefits correlates with the mode of participation, since we find positive income effects only for members selling through the group. A considerable number of members are not participating in collective marketing. Yet, price advantages of collective marketing are small, as highvalue market potentials have not yet been tapped. Increases in income therefore stem from specialization effects and the expansion of banana plantations, facilitated by the improved access to TC planting material. In addition, group members are more likely to follow good agricultural practices in banana cultivation. Hence, beyond prices, farmer groups function as important catalysts for innovation adoption through promoting efficient information flows. In conclusion, some of the motivations for collective action do not seem to apply in the case of banana in Kenya. With significant improvements in road infrastructure, supply chains have become relatively short over the past years. Nor do farmers suffer from monopsonic power of up- and downstream actors that could render collective bargaining beneficial. In terms of price benefits, farmer collective action may thus make more sense in high-value and export markets, which are more difficult to access for individual smallholders. The fac