Understanding carbon dynamics of switchgrass ecosystems is crucial as switchgrass (Panicum virgatum L.) acreage is expanding for cellulosic biofuels. We used eddy covariance system and examined seasonal changes in net ecosystem CO2 exchange (NEE) and its components – gross ecosystem photosynthesis (GEP) and ecosystem respiration (ER) – in response to controlling factors during the second (2011) and third (2012) years of stand establishment in the southern Great Plains of the United States (Chickasha, OK). Larger vapor pressure deficit (VPD > 3 kPa) limited photosynthesis and caused asymmetrical diurnal NEE cycles (substantially higher NEE in the morning hours than in the afternoon at equal light levels). Consequently, rectangular hyperbolic light–response curve (NEE partitioning algorithm) consistently failed to provide good fits at high VPD. Modified rectangular hyperbolic light–VPD response model accounted for the limitation of VPD on photosynthesis and improved the model performance significantly. The maximum monthly average NEE reached up to −33.02 ± 1.96 μmol CO2 m−2 s−1 and the highest daily integrated NEE was −35.89 g CO2 m−2 during peak growth. Although large differences in cumulative seasonal GEP and ER were observed between two seasons, total seasonal ER accounted for about 75% of GEP regardless of the growing season lengths and differences in aboveground biomass production. It suggests that net ecosystem carbon uptake increases with increasing GEP. The ecosystem was a net sink of CO2 during 5–6 months and total seasonal uptakes were −1128 ± 130 and −1796 ± 217 g CO2 m−2 in 2011 and 2012, respectively. In conclusion, our findings suggest that the annual carbon status of a switchgrass ecosystem can be a small sink to small source in this region if carbon loss from biomass harvesting is considered. However, year-round measurements over several years are required to assess a long-term source-sink status of the ecosystem.