The oxygen isotopic composition of cellulose is a valuable proxy in paleoclimate research. However, its application to sedimentary archives is challenging due to extraction and purification of cellulose. Here we present compound-specific δ¹⁸O results of hemicellulose-derived sugar biomarkers determined using gas chromatography-pyrolysis-isotope ratio mass spectrometry, which is a method that overcomes the above-mentioned analytical challenges. The biomarkers were extracted from stem material of different plants (Eucalyptus globulus, Vicia faba and Brassica oleracea) grown in climate chamber experiments under different climatic conditions. The δ¹⁸O values of arabinose and xylose range from 31.4‰ to 45.9‰ and from 28.7‰ to 40.8‰, respectively, and correlate highly significantly with each other (R=0.91, p<0.001). Furthermore, δ¹⁸O_{hemicellulose} (mean of arabinose and xylose) correlate highly significantly with δ¹⁸O_{leaf water} (R=0.66, p<0.001) and significantly with modeled δ¹⁸O_cellulose (R=0.42, p<0.038), as well as with relative air humidity (R=-0.79, p<0.001) and temperature (R=-0.66, p<0.001). These findings confirm that the hemicellulose-derived sugar biomarkers, like cellulose, reflect the oxygen isotopic composition of plant source water altered by climatically controlled evapotranspirative ¹⁸O enrichment of leaf water. While relative air humidity controls most rigorously the evapotranspirative ¹⁸O enrichment, the direct temperature effect is less important. However, temperature can indirectly exert influence via plant physiological reactions, namely by influencing the transpiration rate which affects δ¹⁸O_{leaf water} due to the Péclet effect. In a companion paper (Tuthorn et al., this issue) we demonstrate the applicability of the hemicellulose-derived sugar biomarker δ¹⁸O method to soils and provide evidence from a climate transect study confirming that relative air humidity exerts the dominant control on evapotranspirative ¹⁸O enrichment of leaf water.Finally, we present a conceptual model for the interpretation of δ¹⁸O_{hemicellulose} records and propose that a combined δ¹⁸O_{hemicellulose} and δ²H_n-alkane biomarker approach is promising for disentangling δ¹⁸O_{precipitation} variability from evapotranspirative ¹⁸O enrichment variability in future paleoclimate studies.