Quantitative analysis of wood anatomical characteristics is usually performed using classical microtomy yielding optical micrographs of stained thin sections. It is time-consuming to obtain high quality cross-sections from microtomy, and sections can be damaged. This approach, therefore, is often impractical for those who need quick acquisition of quantitative data on vessel characteristics in wood. This paper reports results of a novel approach using X-ray computed microtomography (microCT) for non-invasive determination of wood anatomy. As a case study, stem wood samples of a 2-year-old beech (Fagus sylvatica L.) and a 3-year-old oak (Quercus robur L.) tree were investigated with this technique, beech being a diffuse-porous and oak a ring-porous tree species. MicroCT allowed non-invasive mapping of 2-D transverse cross-sections of both wood samples with micrometer resolution. Self-developed software 'muCTanalysis' was used for image processing of the 2-D cross-sections in order to automatically determine the inner vessel diameters, the transverse cross-sectional surface area of the vessels, the vessel density and the porosity with computer assistance. Performance of this new software was compared with manual analysis of the same micrographs. The automatically obtained results showed no significant statistical differences compared to the manual measurements. Visual inspection of the microCT slices revealed very good correspondence with the optical micrographs. Statistical analysis confirmed this observation in a more quantitative way, and it was, therefore, argued that anatomical analysis of optical micrographs can be readily substituted by automated use of microCT, and this without loss of accuracy. Furthermore, as an additional application of microCT, the 3-D renderings of the internal microstructure of the xylem vessels for both the beech and the oak sample could be reconstructed, clearly showing the complex nature of vessel networks. It can be concluded that the use of microCT in wood science offers an interesting potential for all those who need quantitative data of wood anatomical characteristics in either the 2-D or the 3-D space. (C) 2004 Elsevier Inc. All rights reserved.