The coffee sub-sector is very important to the Ethiopian economy – in 2005, coffee export generated 41% of foreign exchange earnings – and provides income for approximately 8 million smallholder households. Policy attention to the sector was always considerable, and its importance has been renewed in the latest Poverty Reduction Strategy, the Plan for Accelerated and Sustained Development to End Poverty (PASDEP). PASDEP puts forward a development strategy based on accelerated economic growth, part of which is hoped to be achieved via increased smallholder commercialisation and market integration. This paper addresses commercialisation in selected coffee growing areas in Ethiopia. The objectives of the study were (i) to assess the scale of commercialisation in coffee growing areas and to detect household and farm characteristics which might explain variation in the levels of coffee commercialisation among households; and (ii) to answer two separate questions: why some sampled households didn’t take part in output markets (i.e. identify determinants of market entry) and why some households sold more products than others (i.e. determinants of market supply). Answering these questions will help to identify policy options promoting market participation and commercialisation of smallholder agriculture. Agricultural commercialisation was found to be comparatively high in the studied Weredas (Districts). On the average, farmers marketed 84% of their farm production. Overall, coffee contributed 70% to the total value of output sold. There is, however, a high inter-household differentiation: the 25% highly commercialised smallholders generated over 95% of their cash income from coffee sales, while the bottom 25% earned 63% of their cash income from selling food crops. Keeping other factors constant, the total volume of farm production explained about 72% of the variation in the degree of commercialisation among sampled farm households. Demographic and household factors, wealth and total farm size had no effect on the observed variation in the degree of coffee commercialisation among sampled households. A negative and significant association between the level of household coffee commercialisation and land productivity in non-coffee crops was found, indicating potential trade-offs between the production of coffee, the major cash crop, and other, mainly food crops. No evidence was found of increasing labour intensity as a result of increased coffee production. Similarly, the degree of coffee commercialisation was found to have a statistically insignificant effect on household-level food consumption. Overall, the findings demonstrate the integrated nature of the farming system in coffee growing areas. Despite an overall high level of coffee commercialisation, diversified farming is a strategy pursued by the majority of the surveyed households. The study findings, however, suggest that further specialisation in coffee could enhance overall agricultural commercialisation in the study areas. As the propensity to supply more coffee is significantly higher among households depending more heavily on purchased food, minimising the trade-offs in the production of coffee and non-coffee staple food crops, especially in the short-term, is very important, which signifies the importance of addressing risks associated with food supply and price. In general, increasing smallholder coffee commercialisation is expected to be a viable pathway for agricultural development in coffee growing areas of Ethiopia, if the problem of low productivity, barriers for production expansion (e.g. shortage of farm land or constrained access to farm land) and addressing market risks in both the food and coffee market are addressed by increased research and policy attention.