Recent developments of on-line methods have provided another boost to the determination of stable isotope ratios in organic material. Along with a significant increase in sample throughput, the sample sizes decrease, both of which are necessary conditions to acquire long time series from limited wood amounts. In view of this new technique we reconsidered the most important factors influencing the measured isotopic signature which are (1) pooling, (2) homogeneity, and (3) cellulose extraction. In most cases, pooling (i.e., mixing wood of the same year from different trees) can be made in a simple way by mixing the whole wood available because mass-weighted and unweighted isotope measurements were the same within the error. More attention must be paid in homogenizing the sample. Theoretical considerations underpinned by experimental results suggest a fineness of 0.15 mm (115 mesh) if cellulose is extracted and 0.1 mm (165 mesh) for direct wood analysis. Many of previous studies did not achieve this fineness. We find that wood is as good a climate proxy as cellulose. This is shown by comparing correlations of wood and corresponding cellulose isotope values with meteorological data, which are identical within the uncertainty. ¿ 1998 American Geophysical Union