Lignin has long been suspected to be a major source of stable carbon in soils, notably because of the recalcitrant nature of its polyphenolic structure relative to other families of plant molecules. However, lignin turnover studies have produced conflicting results, most of them suggesting that large proportions of plant-residue lignin decompose within a year of incorporation into soils. Here, we propose a two-reservoir model where lignin in undecomposed plant residue (Lp) can either reach soil fractions where it is somewhat protected from further decomposition (Ls) or is transformed to non-lignin products. Model calibration data were obtained through compound-specific 13C isotopic analyses conducted in a zero- to 9-year chronosequence of maize monoculture after wheat in a temperate loam soil of the Paris basin. Lignin was quantified by CuO oxidation as VSC-lignin, i.e. the sum of vanillil- (V), syringyl- (S) and coumaryl-type (C) phenols. Model calibrations indicate that Lp has a turnover rate faster than 1 year and that 92% is mineralized as CO2 or transformed into other non-lignin products, while only 8% reaches the Ls fraction. Estimated turnover rate of the Ls fraction was 0.05 years-1. The model also suggested that about half of Lp was not measured because it had been excluded from the samples in the process of sieving at 5 mm. In conclusion, the model indicates that chemical recalcitrance alone is not sufficient to explain VSC-lignin turnover in soils, and that, functionally, the most relevant mechanism appears to be the transfer of VSC-lignin molecules and fragments from decomposing plant tissues to soil-protected fractions.