This paper aims at assessing impact of improved upland rice technology on farmers' income in southern Yunnan, China. As technology adoption is not randomly assigned but the result of farmers' self-selection, the paper tackles a methodological issue in assessing the causal effect through propensity-score matching analysis. Employing different matching algorithms and using 473 household survey data in 2000, 2002 and 2004, the study pursues a targeted evaluation that how adoption of improved upland rice technology lets farmers to improve their income. It finds that, estimated by whatever matching algorithm, improved upland rice technology has a robust and positive effect on farmers' income, suggesting that agricultural technology has directly contribution to income increase and poverty alleviation. The effect of technology on income, however, shows a diminishing impact, meaning that as a technology addresses to poverty issue, its effect on income will be gradually lower.