Stable isotope analysis of cellulose is an increasingly important aspect of ecological and palaeoenvironmental research. Since these techniques are very costly, any methodological development which can provide simultaneous measurement of stable carbon and oxygen isotope ratios in cellulose deserves further exploration. A large number (3074) of tree-ring α-cellulose samples are used to compare the stable carbon isotope ratios (δ13C) produced by high-temperature (1400°C) pyrolysis/gas chromatography (GC)/isotope ratio mass spectrometry (IRMS) with those produced by combustion GC/IRMS. Although the two data sets are very strongly correlated, the pyrolysis results display reduced variance and are strongly biased towards the mean. The low carbon isotope ratios of tree-ring cellulose during the last century, reflecting anthropogenic disturbance of atmospheric carbon dioxide, are thus overestimated. The likely explanation is that a proportion of the oxygen atoms are bonding with residual carbon in the reaction chamber to form carbon monoxide. The 'pyrolysis adjustment', proposed here, is based on combusting a stratified sub-sample of the pyrolysis results, across the full range of carbon isotope ratios, and using the paired results to define a regression equation that can be used to adjust all the pyrolysis measurements. In this study, subsamples of 30 combustion measurements produced adjusted chronologies statistically indistinguishable from those produced by combusting every sample. This methodology allows simultaneous measurement of the stable isotopes of carbon and oxygen using high-temperature pyrolysis, reducing the amount of sample required and the analytical costs of measuring them separately. Copyright © 2011 John Wiley & Sons, Ltd.