Coffee and banana are major cash and food crops, respectively, for many smallholders in the East African highlands. Uganda is the largest banana producer and 2nd largest coffee producer in Africa. Both crops are predominantly grown as monocultures. However, coffee–banana intercropping is common in densely populated areas. This study assessed the profitability of intercropped coffee–banana systems compared to mono-cropped systems in regions growing Arabica (Mt. Elgon) and Robusta (south and west) coffee in Uganda. The study was carried out in 152 plots in 2006/2007. Data were collected through structured farmer interviews, field measurements and observations. Coffee yields did not differ significantly (P ⩽ 0.05) between mono-crops and intercrops. Arabica coffee yields were 1.23 and 1.18 t ha−1 year−1 of green beans in mono-cropped and intercropped plots, respectively. Robusta yields averaged 1.25 and 1.09 t ha−1 year−1 of green beans in mono-crops and intercrops, respectively. Banana yields were significantly higher (P ⩽ 0.05) in intercrops (20.19 t ha−1 year−1) compared with mono-crops (14.82 t ha−1 year−1) in Arabica growing region. In Robusta growing region, banana yields were significantly lower (P ⩽ 0.05) in intercrops (8.89 t ha−1 year−1) compared with mono-crops (15.04 t ha−1 year−1). Marginal rate of returns of adding banana to mono-cropped coffee was 911% and 200% in Arabica and Robusta growing regions, respectively. Fluctuations in coffee prices are not likely to affect the acceptability of intercrops when compared with coffee mono-crops in both regions, but an increase in wage rates by 100% can make intercropping unacceptable in Robusta growing region. This study showed that coffee–banana intercropping is much more beneficial than banana or coffee mono-cropping and that agricultural intensification of food and cash crops in African smallholder systems should not solely depend on the mono-crop pathway.