We investigated carbon (C) incorporation and sources of C in the surface CO2 flux at two sites in northern England on peaty (stagnohumic) gley soil, one afforested by Picea sitchensis, the other under continuous Molinia grassland cover. Radiocarbon (C-14) derived from atmospheric nuclear weapons testing was used to trace the incorporation of C into the soil and sources of C in the soil CO2 flux from the soil surface and deeper layers. Larger values of (CO2)-C-14 in surface flux were found at the afforested site (109-110 per cent modern (pM) compared with 107-108 pM at the grassland site). Surface litter fractions (O-i horizon) from the afforested site showed larger C-14 signatures than the equivalent fractions in the grassland (113-115 pM in the forest compared with 106-109 pM in the grassland). Fine root fractions (<2 mm, O-e horizon) had similar signatures at both sites (109 pM in the forest compared with 109-111 pM in the grassland). Humified fractions at 10-cm depth (O-a horizon) showed smaller signatures (100-103 pM) in the forest than the equivalent fraction in the grassland soil (106-114 pM). According to a mixing model that takes into account pool size and C-14 signature, the contributions to surface CO2 fluxes from slow turnover fractions that had resided in the soil for more than one year were greater at the forested site than the grassland site, but contributions from fast-turnover C fixed within the year prior to study showed the opposite trend. The results, taken together with previous work indicating that both site preparation and clear-felling lead to a net loss of C, indicate that long-term fixation in deep soil organic fractions is limited on this soil type under plantation forest over 40-50-year commercial rotations.