The increase of atmospheric carbon dioxide(CO_2) concentration adversely affect several quality traits of rice grains, but the biochemical mechanism remains unclear. The objectives of this study were to determine how changes in the source-sink relationship affected rice quality. Source-sink manipulation was achieved by free-air CO_2 enrichment from tillering to maturity and partial removal of spikelet at anthesis using a japonica rice cultivar Wuyunjing 23. Enrichment with CO_2 decreased the head rice percentage and protein concentration of milled rice, but increased the grain chalkiness. In contrast, spikelet removal resulted in a dramatic increase in the head rice percentage and protein concentration, and much less grain chalkiness. Neither CO_2 enrichment nor spikelet removal affected the starch content, but the distribution of starch granule size showed distinct treatment effects. O n average, spikelet removal decreased the percentage of starch granules of diameter >10 and 5–10 μm by 23.6 and 5.6%, respectively, and increased those with a diameter of 2–5 and <2 μm by 4.6 and 3.3%, respectively. In contrast, CO_2 elevation showed an opposite response: increasing the proportion of large starch granules(>5 μm) and decreasing that of <5 μm. The starch pasting properties were affected by spikelet removal much more than by CO_2 elevation. These results indicated that the protein concentration and starch granule size played a role in chalkiness formation under these experimental conditions. The increase of atmospheric carbon dioxide (CO_2) concentration adversely affect several quality traits of rice grains, but the biochemical mechanism remains unclear. The objectives of this study were to determine how changes in the source-sink relationship affected rice quality. Source-sink manipulation was achieved by free-air CO 2 enrichment from tillering to maturity and partial removal of spikelet at anthesis using a japonica rice cultivar Wuyunjing 23. Enrichment with CO_2 decreased the head rice percentage and protein concentration of milled rice, but increased grain chalkiness. , spikelet removal resulted in a dramatic increase in the head rice percentage and protein concentration, and much less grain chalkiness. Neither CO 2 enrichment nor spikelet removal affected the starch content, but the distribution of starch granule size showed distinct treatment effects. spikelet removal decreased the percentage of starch granules of diameter> 10 and 5-10 m by 23.6 and 5.6%, respectively, and increased those with a diameter of 2-5 and <2 μm by 4.6 and 3.3%, respectively. In contrast, increasing the proportion of large starch granules (> 5 μm) and decreasing that of <5 μm. The results of the protein past and the starch granule size played a role in chalkiness formation under these experimental conditions.