Bundle morphology and adsorptive contributions from nanotubes and impurities are studied both experimentally and by simulation using a computer-aided methodology, which employs a small physisorbed probe molecule to explore the porosity of nanotube samples. Grand canonical Monte Carlo simulation of nitrogen adsorption on localized sites of a bundle is carried out to predict adsorption in its accessible internal pore volume and on its external surface as a function of tube diameter. External adsorption is split into the contributions from the clean surface of the outermost nanotubes of the bundle and from the surface of the impurities. The site-specific isotherms are then combined into a global isotherm for a given sample using knowledge of its tube-diameter distribution obtained by Raman spectroscopy. The structural parameters of the sample, such as the fraction of open-ended nanotubes and the contributions from impurities and nanotube bundles to total external surface area, are determined by fitting the experimental nitrogen adsorption data to the simulated isotherm. The degree of closure between experimental and calculated adsorption isotherms for samples manufactured by two different methods, to provide different nanotube morphology and contamination level, further strengthens the validity and resulting interpretations based on the proposed approach. The average number of nanotubes per bundle and average bundle size, within a sample, are also quantified. The proposed method allows for extrapolation of adsorption properties to conditions where the purification process is 100% effective at removing all impurities and opening access to all intrabundle adsorption sites. (c) 2006 Elsevier Ltd. All rights reserved.