Abstract: Global atmospheric CO₂ concentration has increased from around 280 μmol·mol^-1 in pre-industrial times to the present 400 μmol·mol^-1, and will continue to increase in the future if the emission scenario remains unchanged. As CO₂ is an essential substrate for plant photosynthesis, then, the rising CO₂ has a significant effect on rice production. The late growth stage of rice, from flowering to maturity, is the critical period of yield formation. The source-sink relationship during this reproductive stage plays a crucial role in rice yield formation. Although a lot of research work has been done on regulating the source-sink on rice yield, there are few reports on rice yield response to high CO₂ concentration under regulated sink-source balance. In order to understand how the effect of elevated CO₂ concentration (ambient+200 μmol·mol^-1) on rice yield was influenced by source-sink relationship, we conducted a field experiment in 2014 using a hybrid rice ‘Y Liangyou 2’ and Free Air CO₂ Enrichment (FACE) facility at Yangzhou (119°42′0″E, 32°35′5″N), China. Rice plants were grown under two levels of CO₂ concentration (ambient and elevated) from tillering until maturity. Source-sink manipulation was achieved through cutting off the whole flag leaf (LC1), the top three leaves (LC3) or half of the spikelets at heading (SR, spikelets remove, remove every other primary branch of panicle). Grain yield and its components were investigated. The results showed that under CK conditions (plant intact with no leaf or spikelet cutting), elevated CO₂ concentration increased average rice grain yield by 12% (P < 0.01). This was mainly attributed to slight increase in spikelet number per panicle and grain filling ability. On average, CO₂ elevation increased grain yield by 26% (P < 0.05) and 57% (P < 0.01) for LC1 and LC3 crops, respectively. This response was mainly attributed to the drastic increase in percent grain-filling (LC1: 14%, P < 0.1; LC3: 47%, P = 0.16) and average grain weight (LC1: 11%, P < 0.05; LC3: 24%, P < 0.05). Under SR crops, yield response to elevated CO₂ concentration (+15%, P = 0.01) was similar to that of CK crops. Compared with CK, grain yield under LC1 and LC3 treatments decreased, respectively by 17% (P < 0.01) and 52% (P < 0.01) at heading stage. This decline was mainly due to the decrease in percent grain-filling and average grain weight. Although grain-filling capacity was enhanced by SR treatment, grain yield declined significantly (29%) due to the halved total spikelet number. The response of grain yield to elevated CO₂ concentration was positively correlated with that of percent grain-filling and average grain weight. The results indicated that source-sink manipulation (especially leaf removal) could change the response of grain yield to the elevated CO₂ concentration by affecting rice grain-filling capability at heading stage.