We developed and tested a theoretical model describing carbon isotope discrimination during photosynthesis in tree bark. Bark photosynthesis reduces losses of respired CO2 from the underlying stem. As a consequence, the isotopic composition of source CO2 and the CO2 concentration around the chloroplasts are quite different from those of photosynthesizing leaves. We found three lines of evidence that bark photosynthesis discriminates against 13C. First, in bark of Populus tremuloides, the δ13C of CO2 efflux increased from –24.2‰ in darkness to –15.8‰ in the light. In Pinus monticola, the δ13C of CO2 efflux increased from –27.7‰ in darkness to –10.2‰ in the light. Observed increases in δ13C were generally in good agreement with predictions from the theoretical model. Second, we found that δ13C of dark-respired CO2 decreased following 2–3 h of illumination (P<0.01 for Populus tremuloides, P<0.001 for Pinus monticola). These decreases suggest that refixed photosynthate rapidly mixes into the respiratory substrate pool. Third, a field experiment demonstrated that bark photosynthesis influenced whole-tissue δ13C. Long-term light exclusion caused a localized increase in the δ13C of whole bark and current-year wood in branches of P. monticola (P<0.001 and P<0.0001, respectively). Thus bark photosynthesis was shown to discriminate against 13C and create a pool of photosynthate isotopically lighter than the dark respiratory pool in all three experiments. Failure to account for discrimination during bark photosynthesis could interfere with interpretation of the δ13C in woody tissues or in woody-tissue respiration.