Favorable weather conditions for dryland crop production, including a proper amount of heat and rainfall during the growing season, are critical factors determining yield outcomes. Weather conditions however, are randomly distributed across regions and over time, thus influencing the temporal and geographical patterns of measured crop yield. Failure to account for weather factors when estimating crop yield distributions, time trends or productivity gains can lead to spurious conclusions regarding technology improvement, yield risk and skewness of yield. This paper addresses some limitations in the literature that result from not taking into account weather, and proposes an approach to incorporate weather into modeling yield.