We examine smallholder participation in horticultural markets in Zambia, with two main questions in mind: 1) who participates in horticultural markets? and 2) how does participation affect household income and other welfare outcomes? To control for self-selection bias in the estimation of impacts, we used an endogenous switching framework on nationwide representative data over two agricultural seasons pooled, but controlling for district-level fixed effects. We found that participation is associated with labour availability, farm size, lagged productive assets, social capital through blood kinship links to the chief or headman, level of community participation in the government’s input subsidy programme, and high rainfall variability measured by its coefficient of variation. Participation significantly increased income by 285% overall, increasing to over 300% for femaleheaded households, those cultivating less than one hectare and the extremely poor. These findings provide an empirical foundation to support Zambian policy-makers’ crop-diversification and poverty-reduction agricultural policy objectives.