This study investigates the impact of improved maize varieties on household food security in eastern Zambia using household survey data from a sample of over 800 rural households. Since treatment effect estimates are often prone to misspecification in either the treatment or outcome equation, we use the doubly robust inverse probability weighted regression adjustment method, complemented with propensity score matching on six different food security measures to obtain reliable impact estimates. Generally, we find a positive impact of improved maize adoption on food security across the two econometric approaches. Maize being the most important food staple in Zambia has a great bearing on the food security status of farm households. It is therefore imperative that a conducive environment is created that promotes the adoption of maize yield improving technologies.