This study utilize two-wave household level panel data spanning 5 years on smallholder vegetable producers in Central and Eastern Kenya to assess the effects of commercialization of horticulture on two major poverty outcomes: household income and asset holdings. Methods that exploit panel nature of data to account for observed and unobserved heterogeneity in the sample are used, thus improving upon methods that make no such considerations or rely on cross-sectional data methods. Standards fixed effect, two-step fixed effects approach borrowed from Heckman (1976)’s framework and fixed effect instrumental variable approach find positive effects of commercialization of vegetables through export market pathway on per adult equivalent income. Controlling for heterogeneity and selection bias provides smaller effects of this market pathway compared to the naïve pooled OLS. Similarly, the naïve model overestimates the effect of commercialization through the domestic market pathway on per adult equivalent income. Fixed effect models reveal limited potential of income generated from export market pathway to raise household assets but find positive effect of income from domestic market pathway to improvement of household asset capacity. Results suggest the argument of commercialization of smallholder horticultural farming as “pro-poor” development strategy should look beyond household income to other household welfare aspects such as assets. Further, measuring effects of commercialization of agriculture can be improved by using panel data and addressing both heterogeneity and selection bias, to avoid overestimation of effects of agriculture on poverty. Further research should focus on intra-household distribution and utilization of income generated from the vegetable enterprises.