The aim of this study was to systematically quantify differences in soil carbon and key related soil properties along a replicated land-use intensity gradient on three soil landscapes in northwest New South Wales, Australia. Our results demonstrate consistent land-use effects across all soil types where C, N and C:N ratio were in the order woodland > unimproved pasture = improved pasture > cultivation while bulk density broadly showed the reverse pattern. These land-use effects were largely restricted to the near surface soil layers. Improved pasture was associated with a significant soil acidification, indicating that strategies to increase soil carbon through pasture improvement in these environments might also have associated soil degradation issues. Total soil carbon stocks were significantly larger in woodland soils, across all soil types, compared with the other land-uses studied. Non-wooded systems, however, had statistically similar carbon stocks and this pattern persisted whether or not carbon quantity was corrected for equivalent mass. Our results suggest that conversion from cultivation to pasture in this environment would yield between 0.06 and 0.15 t C/ha/yr which is at the lower end of predicted ranges in Australia and well below values measured in other cooler, wetter environments. We estimate that a 10% conversion rate (cultivation to pasture) across NSW would yield around 0.36 Mt CO2-e/yr which would contribute little to emission reductions in NSW. We conclude that carbon accumulation in agricultural soils in this environment might be more modest than current predictions suggest and that systematically collected, regionally specific data are required for the vegetation communities and full range of land-uses before accurate and reliable predictions of soil carbon change can be made across these extensive landscapes.