Sri Lanka introduced farmer companies (FCs) to link smallholders with preferred markets. Many of these farmer-owned marketing firms failed. The twin objectives of this study are to understand the effect of internal factors - institutional, group and management attributes - on the performance of farmer companies and to make recommendations to improve their performance. The study drew primarily on the New Institutional Economics and management literature to develop a causal model of relationships between the performance of a farmer company and its institutional, group and management attributes. A qualitative, multiple case study research design was used to gather data explaining how shareholders, directors and managers responded to the institutions that characterised their own farmer company. Pairs of successful and failed FCs in each of three core business categories were purposefully selected for in-depth case study. Institutional, group and management attributes were contrasted to test the model’s propositions in a ‘pattern matching’ exercise conducted for each pair of companies. This qualitative analysis identified 34 attributes that affected FC performance. These attributes and seven measures of performance were then subject to hierarchical cluster analysis to triangulate the qualitative findings and to generate more information about relationships between the attributes and performance indicators. The results indicate that FCs are more likely to attract capital and invest in value-adding assets (like brands) when they alleviate the ‘horizon’ problem by making benefits directly proportional to investment. This means that shares should be appreciable and that patrons should pay and receive market-related prices for their inputs and products. Growth in the number of shareholders (outreach) was constrained by a ‘portfolio’ problem in FCs that expected their members to invest equal amounts of equity capital. To avoid this problem, equity shares should be tradable between members and facilitators should help FCs to establish trading platforms. Outreach was also constrained in FCs that arbitrarily limited membership to a small geographical area. Perceptions that external facilitators, executive managers and directors who were not nominated for election to the board by shareholders could influence policy decisions against the interests of majority investors were particularly damaging to investor confidence. Such ‘influence’ problems were also attributed to flawed electoral procedures. All directors should be nominated by shareholders and that voting should be conducted by secret ballot. To improve accountability, the right to hire and fire executive managers should remain with the board of directors, and these executive managers should report to shareholder-directors and not to government agencies. Despite the absence of investment-proportional voting rights, FC performance was not adversely affected by product or shareholder heterogeneity when management remained centralised. Failure to separate company ownership from control undermined company performance by exposing investors, lenders and strategic partners to a severe influence problem. Leaving decisions in the hands of directors and managers does not imply that ordinary shareholders’ views are unimportant when formulating company policy and business strategies. On the contrary, the results suggest that directors and managers should establish forums and processes to elicit the views of shareholders (and other stakeholders) on policy and management issues. Farmer company performance was compromised by the absence of well-defined and regularly observed procedures to develop and implement new strategies, and by inadequate or inappropriate management skills. These management problems may diminish if government facilitators had a clear exit plan from the time the company is established as this would focus their attention on the important task of empowering small farmers to manage a company.