A combination of paired site, time series, and survey approaches were used to estimate the effect of land use change on mineral soil carbon (C), and to identify factors associated with variation. Land-uses compared included podocarp/hardwood forest, improved pasture, and pine plantation. Soil C was significantly related to soil pH that ranged between 3.9–5.9 (0–0.05 m), 3.6–6.0 (0.05–0.10 m), and 4.5–6.1 (0.10–0.50 m) in indigenous forest. Time series data obtained by periodically re-sampling soil (0–0.10 m) in permanent plots in a pine forest previously under pasture showed that mineral soil C decrease by approximately 4 Mg ha⁻¹ by the end of the first rotation. The time series data compared closely with mean results obtained at paired-site throughout New Zealand. Soil C concentration was highly variable in all land-uses, and the evidence suggests that chemical stabilisation of C occurred under acid conditions in native forest, through complexation with Al, and that effects persisted long after conversion of the native forest to other land-uses. The implications of these findings for the design of sampling protocols for soil C are discussed.