The invasion of woody plants into grass-dominated ecosystems has occurred worldwide during the past century with potentially significant impacts on soil organic carbon (SOC) storage, ecosystem carbon sequestration and global climate warming. To date, most studies of tree and shrub encroachment impacts on SOC have been conducted at small scales and results are equivocal. To quantify the effects of woody plant proliferation on SOC at broad spatial scales and to potentially resolve inconsistencies reported from studies conducted at fine spatial scales, information regarding spatial variability and uncertainty of SOC is essential. We used sequential indicator simulation (SIS) to quantify spatial uncertainty of SOC in a grassland undergoing shrub encroachment in the Southern Great Plains, USA. Results showed that both SOC pool size and its spatial uncertainty increased with the development of woody communities in grasslands. Higher uncertainty of SOC in new shrub-dominated communities may be the result of their relatively recent development, their more complex above- and belowground architecture, stronger within-community gradients, and a greater degree of faunal disturbance. Simulations of alternative sampling designs demonstrated the effects of spatial uncertainty on the accuracy of SOC estimates and enabled us to evaluate the efficiency of sampling strategies aimed at quantifying landscape-scale SOC pools. An approach combining stratified random sampling with unequal point densities and transect sampling of landscape elements exhibiting strong internal gradients yielded the best estimates. Complete random sampling was less effective and required much higher sampling densities. Results provide novel insights into spatial uncertainty of SOC and its effects on estimates of carbon sequestration in terrestrial ecosystem and suggest effective protocol for the estimating of soil attributes in landscapes with complex vegetation patterns.