The carbon (C) budget of managed grassland in a cool-temperate region of Japan was estimated using a combination of eddy covariance and the biometric method for five years, to evaluate the effect of manure application. Chemical fertilizer was applied to the fertilizer (F) plot at a rate of 79 ± 20 kg N ha−1 yr−1. In the manure (M) plot, dairy cattle manure was applied at a rate of 10 Mg fresh matter ha−1 yr−1 (1923 ± 407 kg C ha−1 yr−1, 159 ± 68 kg N ha−1 yr−1). There was no significant difference in seasonal gross primary production (GPP) and harvest between the treatment plots, indicating that both fertilizer and manure can increase the biomass production. Annual net ecosystem production (NEP) and ecosystem respiration (RE) was significantly different between the treatment plots. The difference in RE, and between M and F plots approximates heterotrophic respiration of manure (RHm), which ranged from 0.9 to 1.3 Mg C ha−1 yr−1. Average annual RHm was 1.1 ± 0.4 Mg C ha−1 yr−1, and accounted for 56% of the total amount of applied manure C. The annual net biome production (NBP) in the M plot (from 0.0 to 1.5 Mg C ha−1 yr−1) was significantly higher than in the F plot (−1.4 to 0.5 Mg C ha−1 yr−1). The long-term effect of manure application combined with chemical fertilizer did not reduce grass production compared with chemical fertilizer only; however, manure application decreased the NEP throughout manure decomposition, and long-term manure application enhanced the NBP.