This paper analyzes the adoption and welfare impacts of improved cassava varieties in Zambia using data from a sample of 500 farm households. Using different treatment effect estimators—endogenous switching regression, propensity score matching, and inverse probability weighting, the paper shows that adoption of improved cassava varieties leads to significant gains in crop yields, household income, and food security. Results further show that improved cassava varieties have significant poverty-reducing impacts in Zambia. Stimulating agricultural growth largely depends on policies that stimulate adoption of improved cassava varieties.