Soil carbon stock change between two major land uses in New Zealand was measured by sampling paired plots across the boundaries of low productivity grassland and forest planted pre-1990. The national soil carbon monitoring system uses low productivity grassland as a benchmark to evaluate soil carbon stock change for other land uses. The goal was to validate earlier estimates of the effect of pre-1990 afforestation and to reduce their level of uncertainty. We selected a set of sites to represent the national stocks of forests planted pre-1990. Previous studies derived estimates of the land-use effect on soil carbon for afforestation ranging from +1.6 to −8.5 t/ha to 30 cm depth. For all estimates, the 95% confidence interval spanned zero. Our study used nine of the previous paired-plot sites and sampled and analysed 21 new sites. The land-use effect of change from grassland to forest planted pre-1990 was estimated at −17.4 t/ha. The 95% confidence interval ranged from −10.1 to −24.6 t/ha and did not include zero change. The result supported the soil carbon monitoring system assumption that forests planted pre-1990 have significantly lower soil carbon stocks than the low-productivity-grassland standard. Evidence of stock change occurred in depth increments to 0.2 m but with no significant change for the 0.2–0.3 m increment. This suggests that the sampling depth of 0.3 m was adequate for the estimation of soil carbon stock change.