Abstract. Temporal dynamics of C isotopic composition (δ¹³C) of CO₂ and leaf litter was monitored during a litter decomposition experiment using Arbutus unedo L., as a slow decomposing model substrate. This allowed us (1) to quantify isotopic discrimination variation during litter decomposition, and (2) to test whether selective substrate use or kinetic fractionation could explain the observed isotopic discrimination. Total cumulative CO₂-C loss (C_L) comprised 27% of initial litter C. Temporal evolution of C_L was simulated following a three-C-pool model. Isotopic composition of respired CO₂ (δ_RL) was higher with respect to that of the bulk litter. The isotopic discrimination Δ_(L/R) varied from −2‰ to 0‰ and it is mostly attributed to the variations of δ_RL. A three-pool model, with the three pools differing in their δ¹³C, described well the dynamic of Δ_(L/R), in the intermediate stage of the process. This suggests that the observed isotopic discrimination between respired CO₂ and bulk litter is in good agreement with the hypothesis of successive consumption of C compounds differing in δ¹³C during decomposition. However, to explain also ¹³C-CO₂ dynamics at the beginning and end of the incubation the model had to be modified, with discrimination factors ranging from −1‰ to −4.6‰ attributed to the labile and the recalcitrance pool, respectively. We propose that this discrimination is also the result of further selective use of specific substrates within the two pools, likely being both the labile and recalcitrant pool of composite nature. In fact, the 2‰ ¹³C enrichment of the α-cellulose observed by the end of the experiment, and potentially attributable to kinetic fractionation, could not explain the measured Δ_(L/R) dynamics.