Poverty among rural farming households in sub-Saharan Africa (SSA) is associated with low adoption of modern farming technologies, especially improved crop varieties. Most studies that investigated adoption and impacts of improved crop varieties in SSA are based on farmers’ selfreported adoption status and average treatment effects. However, farmers self-reported adoption status is susceptible to errors and assessing adoption impacts using average treatment effects do not account for farmers’ heterogeneity. To address these challenges, we used DNA-fingerprinting data and Marginal Treatment Effect (MTE) framework to analyze adoption and impacts of adopting improved groundnut varieties (IGVs) in Northern Nigeria. DNA-fingerprinting results showed 57% adoption rate compared to 45% self-reported by farmers. About 29% of the sampled farmers made type I error (mistaking local varieties for improved varieties) while 44% made type II error (mistaking improved varieties for local varieties). Formal sources of seed information and empowering agricultural extension to reach more farmers was significant in ensuring accurate variety identification. Further, adopting IGV significantly reduced poverty gap and poverty severity, especially among households headed by females, older persons, lowly educated and those with limited access to credit. Therefore, policy options that enhance adoption of IGVs will significantly help in reducing poverty.