Rationale The carbon isotope (δ13C value) composition of fossil plant material is routinely used as a proxy of past climate and environment change. However, palaeoclimate interpretation requires assumptions about the stability of δ13C values in plant material during its decomposition and incorporation into sediments. Previous work on modern angiosperm species shows δ13C changes of several per mille during simulated decomposition experiments. However, no such tests have been undertaken on non-flowering plants, which are found extensively within the geological record. These plants have distinctly different cellulose-to-lignin ratios from those of their angiosperm counterparts, potentially creating hitherto unknown variations in the original to fossil δ13C signatures. Methods To test the extent of δ13C change during decomposition we have subjected a number of plants, representing more basal, non-flowering plant lineages (cycads, ferns, horsetails and dawn redwood), to artificial decay using a hydrothermal maturation technique at two temperatures over periods of up to 273 hours. Subsamples were extracted every 12–16 hours and analysed for their δ13C and %C values using a Carlo Erba 1500 elemental analyser, and VG TripleTrap and Optima mass spectrometers. Results The %C values increased for all samples through the maturation process at both temperatures with the largest increases observed within the first 24 hours. Decreases in δ13C values were observed for all plants at 300°C and for two of the species at the lower temperature (200°C). The maximum shift in the δ13C value related to experimental decomposition was −0.90‰ (horsetail), indicating a preferential loss of 13C during thermal maturation. Conclusions The reduction in the δ13C value potentially suggests a preferential loss of isotopically heavier cellulose in relation to the isotopically lighter lignin component during maturation. The isotopic offset observed here (