Despite the major role of climate in agricultural production, few studies have analyzed how weather fluctuations affect the measurement and decomposition of Total Factor Productivity (TFP). This article proposes a novel framework to decompose TFP change accounting for the influence of weather. Specifically, we estimate the contribution of weather variations, technical change, technical and allocative efficiency, as well as markup, scale and price effects to TFP change. The underlying technology is represented by a multi-input, multi-output flexible input distance function with quasi-fixed inputs of production, and is estimated for major U.S. producing regions using Bayesian methods. To assess the role of weather in the decomposition of TFP growth, we contrast findings from our proposed method with those of a baseline model that ignores weather effects. Overall, our TFP growth estimates are highly similar to those obtained from official USDA indices. However, we find that the contribution of non-weather components to TFP is 14% lower when we account for weather variations. This weatherrelated bias is particularly strong in the Central region of the country. This overestimation of TFP growth that is attributable to non-weather components in previous research thus implies that estimated rates of return to public R&D are also overestimated, which has profound policy implications. This is the first study to document how ignoring weather can bias the decomposition of TFP change estimates.