Evaluating agricultural extension programs enables the assessment of success and continuity using various research designs, data, and methodologies. We synthesize the literature on the impact of agricultural extension services to compare the magnitude and direction across four outcomes: agricultural production, input use, adoption, and welfare of farming households. We also examine their variation across the study attributes. Our literature search yielded 120 causal inference studies, which produced 579 estimates published between 2004 and 2025. We then employed meta-regression analysis for empirical analysis. Our results show that the estimated impact of agricultural extension services reported in the literature increased over time across all estimates and outcomes related to agricultural production, input use, and household welfare, while it decreased over time on outcomes associated with adopting agricultural technologies. Other results indicate that the average estimate of the impact of agricultural extension services reported in the literature is positive and statistically significant, with small but consistent effects across all outcome domains, as revealed through meta-regression and bias-corrected models. We also find that some study attributes are associated with the variation in the study's reported impact of agricultural extension services.
