The main objective of this study was to evaluate the impact of adopting improved maize varieties on crop yield in Uganda using propensity score matching (PSM) algorithms to control for hidden selection bias. The study employed cross-sectional household data collected in 2005/2006 by the Uganda Bureau of Statistics (UBOS) across the country. Data were divided into two parts; the full sample and smallholder farmer sub-sample (those that cultivated less than 5 acres or 2 Hectares of maize in 2004 and 2005). Then estimation was made of the difference in yields between the adopters and non adopters, the average treatment effect on the treated (ATT) for the full sample and smallholder farmer sub-sample. Matching methods were used to control for hidden selection bias and the standardized bias measure was used to check for the quality of matching. The results indicate that adoption of improved maize seed had a robust positive and significant effect on yields obtained by farmers. The results were consistent for both the full and smallholder farmer sub-samples. Sensitivity analysis using Rosenbaum bounds indicated that the ATT results were fairly robust to hidden selection bias. That is, the results were not sensitive to unobserved selection bias. Therefore it is confidently asserted that the estimated average difference in maize yields between the adopters and non-adopters was due to the effect of planting improved maize seed.