Assessment of soil carbon (C) stock changes over time is typically based on the application of two methods, namely (i) repeated soil inventory and (ii) determination of the ecosystem C budget or net biome productivity (NBP) by continuous measurement of CO2 exchange in combination with quantification of other C imports and exports. Here, we applied both methods in parallel to determine C stock changes of two temperate grassland fields previously converted from long-term cropland. The grasslands differed in management intensity with either intensive management (high fertilization, frequent cutting) or extensive management (no fertilization, less frequent cutting). Soil organic C stocks (0–45 cm depth) were quantified at the beginning (2001) and the end (2006) of a 5 year observational period using the equivalent soil mass approach. For the same period and in both fields, NBP was quantified from net CO2 fluxes monitored using eddy covariance systems, and measured C import by organic fertilizer and C export by harvest. Both NBP and repeated soil inventories revealed a consistent and significant difference between management systems of 170 ± 48 and 253 ± 182 g C m−2 a−1, respectively. For both fields, the inventory method showed a tendency towards higher C loss/smaller C gain than NBP. In the extensive field, a significant C loss was observed by the inventory but not by the NBP approach. Thus neither flux measurements nor repeated soil sampling may be suitable for tracking absolute changes in SOC, but both give similar answers with respect to relative changes.