AbstractTree stem respiration (R_S) is a substantial component of the forest carbon balance. The mass balance approach uses stem CO₂ efflux and internal xylem fluxes to sum up R_S, while the oxygen‐based method assumes O₂ influx as a proxy of R_S. So far, both approaches have yielded inconsistent results regarding the fate of respired CO₂ in tree stems, a major challenge for quantifying forest carbon dynamics. We collected a data set of CO₂ efflux, O₂ influx, xylem CO₂ concentration, sap flow, sap pH, stem temperature, nonstructural carbohydrates concentration and potential phosphoenolpyruvate carboxylase (PEPC) capacity on mature beech trees to identify the sources of differences between approaches. The ratio of CO₂ efflux to O₂ influx was consistently below unity (0.7) along a 3‐m vertical gradient, but internal fluxes did not bridge the gap between influx and efflux, nor did we find evidence for changes in respiratory substrate use. PEPC capacity was comparable with that previously reported in green current‐year twigs. Although we could not reconcile differences between approaches, results shed light on the uncertain fate of CO₂ respired by parenchyma cells across the sapwood. Unexpected high values of PEPC capacity highlight its potential relevance as a mechanism of local CO₂ removal, which merits further research.