The objective of this study is to clarify the carbon dynamics and the effect of fertilizer or manure management on the carbon budget in a managed grassland of reed canary grass in Southern Hokkaido, Japan. The fertilizer, manure and control plots were established in the grassland, and carbon budgets were estimated for 2 years using an ecological technique. In the manure plot, beef cattle manure was applied at the rate of 43–44 Mg fresh matter ha −1 year −1, and a supplement of chemical fertilizer was also added to equalize the application rate of mineral nitrogen (N) to that of the fertilizer plots (164–184 kg N ha −1 year −1). The aboveground net primary production (ANPP) was significantly larger in the fertilizer and manure plots than in the control plot. The belowground biomass indicated no distinct seasonal change; therefore the belowground net primary production (BNPP) was supposed to be zero. The CO 2 fluxes from soil surface including roots in the fertilizer, manure and control plots were higher in spring (beginning of March–end of June) than in other seasons (beginning of July–end of February) at the same soil temperature, while the relationship between CO 2 flux and soil temperature in the root-excluded plot was well explained by a single exponential regression throughout the study period. This suggests that there might be an increment of root respiration (RR) or heterotrophic respiration of litter (RHl) produced by fine roots (which means BNPP = RHl). Heterotrophic respiration of soil (RHs) ranged from 3.7 to 4.3 Mg C ha −1 year −1. Heterotrophic respiration of manure (RHm) accounted for 41% in the added carbon in manure for 2 years. Finally, the net ecosystem production (NEP) in this study was estimated as ANPP–RHs–RHm. The annual NEP were 2.6 and 1.3 Mg C ha −1 year −1 for the fertilizer plot, −1.6 and −1.0 Mg C ha −1 year −1 for the manure plot, and −1.1 and −0.8 Mg C ha −1 year −1 for the control plot in 2005 and 2006, respectively. The annual net biome production (NBP) in the fertilizer and control plots ranged from −3.9 to −2.9 Mg C ha −1 year −1, suggesting the loss of carbon in the field. On the other hand, NBP in the manure plot was 0.3 and 0.0 Mg C ha −1 year −1 in 2005 and 2006, respectively. These results indicate that manure application is necessary to prevent carbon loss in the managed grassland.