Bioremediation of soil contaminated by organic compounds can remove the contaminants to a large extent, but residual contamination levels may remain which are not or only slowly biodegraded. Residual levels often exceed existing clean-up guidelines and thereby limit the use of bioremediation in site clean-up. A method for estimating the expected residual levels would be a useful tool in the assessment of the feasability of bioremediation. In this study, three soil types from a creosote-contaminated field site, which had been subjected to 6 months of bioremediation in laboratory column studies, were used to characterize the residual contamination levels and assess their availability for biodegradation. The soils covered a wide range of organic carbon levels and particle size distributions. Results from the biodegradation studies were compared with desorption rate measurements and selective extractability using butanol. Residual levels of polycyclic aromatic hydrocarbons after bioremediation were found to be strongly dependent on soil type. The presence of both soil organic matter and asphaltic compounds in the soil was found to be associated with higher residual levels. Good agreement was found between the biodegradable fraction and the rapidly desorbable fraction in two of the three soils studied. Butanol extraction was found to be a useful method for roughly estimating the biodegradable fraction in the soil samples. The results indicate that both desorption and selective extraction measurements could aid the assessment of the feasability for bioremediation and identifying acceptable end-points.