PurposeThe objective of this study is to estimate the impacts of access to irrigation on farm income, household income and income diversification.Design/methodology/approachThis study employs an endogenous switching regression (ESR) model to address the selection bias arising from both observed and unobserved factors and analyze cross-sectional data collected from Fujian, Henan and Sichuan provinces in China. The authors use the Simpson index to measure household income diversification. The propensity score matching (PSM) model and control function approach are also used for comparison purpose.FindingsAfter controlling for the selection bias, the authors find that access to irrigation has a positive and statistically significant impact on rural incomes and diversification. The treatment effects of access to irrigation are to increase farm income, household income and income diversification by around 14, 10 and 107%, respectively. The positive effects of access to irrigation are confirmed by the estimates of the PSM model and control function approach. Further analysis reveals that the irrigation effects on rural incomes and diversification are heterogeneous between small-scale and large-scale farmers and between male-headed and female-headed households.Practical implicationsThe authors’ findings suggest that the government should continue to improve irrigation infrastructure construction in rural China to promote smallholder farmers' water access and at the same time facilitate farmers' access to better agronomic and irrigation information. There exist gender and farm size related income and diversification effects of access to irrigation, and the irrigation access is associated with farm location. Thus, when developing regional irrigation programs consideration needs to be taken of whether the rural farming systems are dominated by male/female household heads and land fragmentation/consolidation issues.Originality/valueAlthough a large body of literature has investigated the effects of irrigation development in rural areas, little is known about the impact of access to irrigation on income diversification. The selection bias associated with unobserved heterogeneities is usually neglected in previous studies. This study provides the first attempt by examining the impacts of access to irrigation on rural incomes and diversification, using the ESR model to address both observed and unobserved selection bias.