Long-term field measurements of the exchange of CO₂ between terrestrial ecosystems and the atmosphere can contribute to our knowledge on the role of a particular ecosystem in the global carbon cycle. To date, little information is available on the carbon fluxes in a typical steppitied desert, especially in areas with annual precipitation less than 200 mm. The community dominated by Reaumuria soongorica Maxim. and Salsola passerina Bge. is one of the most common vegetation types in desert areas in China. In this program, the R.soongorica-S.passerina community in the Tennger Desert was selected as our study object. The diurnal variation of the net ecosystem CO₂ exchange (NEE), the gross primary productivity (GPP), the ecosystem respiration (Reco) and the soil respiration (Rsoil) were measured using the closed static chamber method. The results obtanined by the chamber method were compared with eddy covariance measurements and the main results were as follows: (1) NEE of the R.soongorica-S.passerina community exhibited a negative peak (negative values indicate CO₂ uptake by the desert ecosystem) in the morning between 6:00-9:00, and then increased and showed positive values (positive values indicate CO₂ release from the desert ecosystem to atmosphere) with the largest efflux peak occurring between 12:00-15:00. The diurnal variation of NEE in the R.soongorica-S.passerina community was similar to the S. passerina population, which means the variation of the R. soongorica-S. passerina community was determined by the S. passerina population. The mean NEEs of the R.soongorica population, S.passerina population and R.soongorica-S.passerina community were 0.018, 0.020 and 0.028 mgCO₂ m^-2 s^-1, respectively. (2) The diurnal variation of the respiration of the R. soongorica population, S. passerina population, soil and the whole community were similar, and exhibiting single-peak curves at daytime with the peak appearing at between 12:00-15:00, and the mean values of respiration of R. soongorica population, S. passerina population, soil and the whole community were 0.121, 0.062, 0.029, and 0.040 mgCO₂ m^-2 s^-1, respectively. We calibrated the carbon efflux using the coverage-weighted-average method and found that the R. soongorica population, S. passerina population and the soil system comprise 9%, 21% and 70% of the total R.soongorica-S.passerina community ecosystem respiration, respectively. Therefore the ecosystem respiration derived mainly from the soil system. (3) The variation of NEE monitored using the closed static chamber method was generally consistent with that by using eddy covariance method. The NEE measured by the closed static chamber was linearly correlated with the eddy covariance, and the correlative value was 0.7. The NEE measured by the closed static chamber was higher than the eddy covariance, with the mean values were 0.028 mgCO₂ m^-2 s^-1 (the former), 0.015 mg CO₂ m^-2 s^-1 (the latter), respectively. The result measured by the eddy covariance vs closed static chamber was 0.54. In conclusion, the carbon efflux of desert ecosystem maybe overestimated by using the closed static chamber method.