Arid and semiarid ecosystems make up approximately 41% of Earth’s terrestrial surface and are suggested to regulate the trend and interannual variability of the global terrestrial carbon (C) sink. Biological soil crusts (biocrusts) are common dryland soil surface communities of bryophytes, lichens, and/or cyanobacteria that bind the soil surface together and that may play an important role in regulating the climatic sensitivity of the dryland C cycle. Major uncertainties exist in our understanding of the interacting effects of changing temperature and moisture on CO2 uptake (photosynthesis) and loss (respiration) from biocrust and sub-crust soil, particularly as related to biocrust successional state. Here, we used a mesocosm approach to assess how biocrust successional states related to climate treatments. We subjected bare soil (Bare), early successional lightly pigmented cyanobacterial biocrust (Early), and late successional darkly pigmented moss-lichen biocrust (Late) to either ambient or + 5°C above ambient soil temperature for 84 days. Under ambient temperatures, Late biocrust mesocosms showed frequent net uptake of CO2, whereas Bare soil, Early biocrust, and warmed Late biocrust mesocosms mostly lost CO2 to the atmosphere. The inhibiting effect of warming on CO2 exchange was a result of accelerated drying of biocrust and soil. We used these data to parameterize, via Bayesian methods, a model of ecosystem CO2 fluxes, and evaluated the model with data from an autochamber CO2 system at our field site on the Colorado Plateau in SE Utah. In the context of the field experiment, the data underscore the negative effect of warming on fluxes both biocrust CO2 uptake and loss—which, because biocrusts are a dominant land cover type in this ecosystem, may extend to ecosystem-scale C cycling.