Intensification of grazed grasslands following conversion from dryland to irrigated farming has the potential to alter ecosystem carbon (C) cycling and affect components of carbon dioxide (CO 2 ) exchange that could lead to either net accumulation or loss of soil C. While there are many studies on the effect of water availability on biomass production and soil C stocks, much less is known about the effect of the frequency of water inputs on the components of CO 2 exchange. We grew Bermuda grass ( Cynodon dactylon L.) in mesocosms under irrigation frequencies of every day (I 1 treatment, 30 d), every two days (I 2 treatment, 12 d), every three days (I 3 treatment, 30 d), and every six days (I 6 treatment, 18 d, after I 2 treatment). Rates of CO 2 exchange for estimating net ecosystem CO 2 exchange ( F N ), ecosystem respiration ( R E ), and soil respiration ( R S ) were measured, and gross C uptake by plants ( F G ) and respiration from leaves ( R L ) were calculated during two periods, 1–12 and 13–30 d, of the 30-d experiment. During the first 12 d, there were no significant differences in cumulative F N (mean ± standard deviation, 61 ± 30 g C m -2 , n = 4). During the subsequent 18 d, cumulative F N decreased with decreasing irrigation frequency and increasing cumulative soil water deficit ( W ), with values of 70 ± 22, 60 ± 16, and 18 ± 12 g C m -2 for the I 1 , I 3 , and I 6 treatments, respectively. There were similar decreases in F G , R E and R L with increasing W , but differences in R S were not significant. Use of the C 4 grass growing in a C 3 -derived soil enabled partitioning of R S into its autotrophic ( R A ) and heterotrophic ( R H ) components using a 13 C natural abundance isotopic technique at the end of the experiment when differences in cumulative W between the treatments were the greatest. The values of R H and its percentage contributions to R S (43% ± 8%, 42% ± 8%, and 8% ± 5% for the I 1 , I 3 , and I 6 treatments, respectively) suggested that R H remained unaffected across a wide range of W and then decreased under extreme W . There were no significant differences in aboveground biomass between the treatments. Nitrous oxide (N 2 O) emission was measured to determine if there was a trade-off effect between irrigation frequency and increasing W on net greenhouse gas emission, but no significant differences were found between the treatments. These findings suggest that over short periods in well-drained soil, irrigation frequency could be managed to manipulate soil water deficit in order to reduce net belowground respiratory C losses, particularly those from the microbial decomposition of soil organic matter, with no significant effect on biomass production and N 2 O emission.