While several studies have linked rural welfare with non-farm employment, the available evidence suggests that whether participation in non-farm employment is welfare improving is intrinsically an empirical question. In addition, the existing literature places emphasis on food and total household expenditure such that there is a dearth of evidence as far as the role of non-food expenditure is concerned. Using household survey data from Burkina Faso, this paper shows that the extent of the impact varies with types of non-farm employment and household expenditure. The results show that participation in wage employment does not affect any type of expenditure. Non-farm self-employment during the dry season increases non-food expenditure but has no significant effect on food expenditure. Non-farm self-employment during both dry and cropping seasons, and total non-farm employment are associated with higher food and non-food expenditures, with higher impact on nonfood expenditure. The general pattern of the results suggests that smallholder farmers tend to classify agriculture as the primary source of food while incomes from non-farm sources are preferably oriented towards non-food consumption. From a policy perspective, this shows that promoting the rural non-farm economy will improve welfare in the semi-arid areas. Potential areas of interventions include (i) implementation of the minimum guarantee wage scheme to improve earnings from wage employment and, (ii) development of a vibrant rural credit scheme to improve access to non-farm self-employment.☆ We thank the handling Editor and anonymous referees for their helpful thoughts, comments, and guidance. We are grateful to Chris Barrett, Kimseyinga Savadogo, Elis e Ou edraogo and Yiribo Bambio for facilitating data access, and Ramatu Al-Hassan for assistance in reading and editing. We are also grateful to Agricultural & Applied Economics seminar participants at the