Rural non-agricultural employment (RNAE) is being increasingly emphasized as a potential pathway out of rural poverty for people who are unable to secure their income in agriculture. Although average earnings in the rural non-agricultural sector are higher than in agriculture, it is unclear whether income prospects are systematically better in non-agricultural activities than in agriculture. This paper tests for existence of earnings differentials between agricultural and rural non-agricultural employment, while controlling for worker and household characteristics. A theoretical farm household model is proposed that predicts that there will be no sectoral earnings differential for unskilled labor, whereas skilled labor will be better off in the non-agricultural sector. Based on Peruvian household data, the empirical findings do not support the notion that unskilled workers would earn a higher income by switching from agriculture to RNAE. Instead it tends to be the relatively well educated who might benefit from higher returns to education in RNAE than in agriculture, consistent with the predictions of the theoretical model.