Simulation experiments were undertaken to analyze the difference ofLarrea tridentata desert ecosystem in Phoenix, USA responses to elevated atmospheric CO₂ concentration between dry and wet years. A physiologicalbased ecosystem model, PALS-FT was used in the simulation analysis. Aboveground net primary productivity （ANPP） and annual accumulation rate of soil organic matter （SOM） increased nonlinearly with increasing [CO₂] in wet years, but linearly in normal and dry years, while N content in soil （N_soil） decreased nonlinearly with increasing [CO₂] in all types of years. The absolute changes of ANPP and SOM were always larger in wet years than in normal and dry years, and the relative changes of them were dependent greater on the [CO₂] level of interest rather than on year types （i.e., wet, normal, and dry years）. However, both the absolute and relative changes of Nsoil were always larger in wet years than in normal and dry years. The absolute changes of ANPP in different plant functional types （PFTs） responding to elevated [CO₂] were larger in wet years than in normal and dry years, but the relative changes varied also PFT-specifically. Shrub and subshrub FTs showed larger responses in dry years, whereas C₃ and C₄ annuals showed larger responses in wet years. Therefore, responses of desert ecosystems to a future elevated CO₂ would be markedly influenced by precipitation patterns.