Timescales of soil organic carbon (SOC) turnover in forests were investigated with soil radiocarbon data. The 12C/14C ratios were measured by accelerated mass spectroscopy on soil sampled from a deciduous temperate forest in Switzerland during 1969-1995. The resulting Delta14C values (125-1740/00) were in line with previously published 14C soil data. We applied FORCLIM-D, a model of nonliving organic matter decomposition including nine litter and two soil compartments to estimate SOC turnover times for this forest type. Carbon 14 aging in woody vegetation was explicitly accounted for. Parameters were calibrated to match radiocarbon ratios observed for forest soils at Meathop Wood, United Kingdom [Harkness et al., 1986]. We estimated that roughly 50-94% (best estimate, 49%) of foliar litter carbon and 11-74% (73%) of fine root litter carbon are eventually respired as CO2 at Meathop Wood; the rest is transferred to soil humus, where it undergoes further decomposition. Turnover times for the 0-20 cm mineral soil layer ranged from 9-50 years (25 years) for a fast overturning soil compartment comprising 38-74% (68%) of bulk SOC and 155-10,018 years (3,570 years) for a slowly overturning compartment. For the Swiss site, SOC turnover times were in the same range. Parameter uncertainties were correlated and induced by uncertainties in 14C observations from small-scale spatial inhomogeneities, sample preparation and by lack of reliable 14C observations for the ``prebomb'' test period. Model-based estimates of soil organic C turnover derived from 14C data must be used cautiously since they depend on the underlying model structure: bypassing litter in FORCLIM-D overestimated SOC turnover by a factor of 2.5. Such an error might remain undetected in studies lacking samples from the late 1960s and early 1970s. Thus litter C turnover should be included when estimating SOC turnover in temperate forests from 14C data.