We tested a grain production model experimentally under field conditions. The model is a simple, mathematical expression that describes the increase of grain weight (ΔG) as a function of the total Drymatter production (ΔW). It has two parameters: grain growth potential (ΔGP) and the maximum mobilizable reserves of the non-grain plant parts (SP). Both of these parameters are to be determined experimentally. The model is originally developed for the relation between the increase of grain (i.e., yield) and the total Drymatter production of a single plant over a whole grain-filling period. Two rice cultivars, Nipponbare and Milyang 23, were grown in an irrigated field. Both cultivars were subjected to thinning and shading treatments after heading, so that they were under six radiation levels during the grain -filling period. Plants were harvested at 0 day, 15 days, 30 days and 45 days after heading to determine the responses for each 15 days in the early, the middle and the late grain-filling stage. The model predictions fit generally well the experimentally determined ΔG-ΔW relations, although the results for the late grain-filling stage were less satisfactory. We therefore conclude that our model is valid for a shorter term response of rice population. The model seems to be useful as a component of a mechanistic crop production model. © 1990, CROP SCIENCE SOCIETY OF JAPAN. All rights reserved.