Pools and fluxes of carbon in ecosystems are regulated by climate and its interaction with time, soils, topography, plants, animals and disturbance. Changes in land cover have a direct impact on the carbon cycle via effects on primary production and litter quality; and indirect effects associated with modification of microclimate and rates of erosion and deposition. Encroachment of woody plants into grasslands has caused extensive land cover change in drylands. However, robust generalizations regarding the consequences of this vegetation change for the carbon cycle have yet to emerge. Studies quantifying the direction of change in soil organic carbon (SOC) pools range from positive to neutral to negative; and where SOC accumulation has been recorded, rates vary by a factor of 10. These contrasting responses will be examined with the aim of identifying underlying processes that determine how shrubs alter the carbon cycle subsequent to their establishment in grasslands. It will be argued that knowledge of such processes will help resolve current controversies and contradictions. With greenhouse gas mitigation and carbon sequestration issues gaining national and international prominence, biogeochemical and land surface-atmosphere perspectives suggest the need for new approaches for evaluating the functional role of woody plants in rangelands. Carbon-accounting perspectives introduce cost/benefit trade-offs that have not been traditionally considered when assessing vegetation management on rangelands. It will be argued that perverse incentives may develop under carbon sequestration subsidies whereby societies may become increasingly willing to tolerate or even subsidize land use practices that promote proliferation of woody plants historically regarded as undesirable. Such scenarios would have adverse impacts on the conservation of grassland ecosystems and the plants and animals endemic to them.