It is often recommended that consumption of dietary fibre should in increased to protect against colorectal cancer. However, although more than 95% of dietary fibre is contributed by whole plant cell walls, very little experimental work has been done using whole plant cell walls. These may protect by adsorbing carcinogens, thus lowering their effective concentration in the alimentary tract, and by carrying the carcinogens out of the body in the faeces. However, plant cell walls vary widely in their composition and physical properties, and not all cell walls will necessarily have protective properties. We therefore isolated 4 plant cell-wall preparations with contrasting compositions as models of the types of cell walls that occur in the diet. We investigated the abilities of these preparations to adsorb in vitro 6 heterocyclic aromatic amines (HAAs). HAAs occur in the human diet and several are colon carcinogens, at least in rats. We found that the ability of the HAAs to adsorb to the plant cell walls increased with increasing hydrophobicity of the HAA, measured as the calculated logarithm of the partition coefficient between 1-octanol and water (C logP). A cell-wall preparation containing mainly the walls of parenchyma cells (the most common cell type in food plants) had only poor adsorptive ability. A cell-wall preparation from commercial cork had the best adsorptive ability. This preparation was the most hydrophobic of those examined because the cell walls contained the polymer, suberin, together with associated waxes. The preparation modelled suberized cell walls which occur in the diet, for example in potato skins. The other two cell-wall preparations contained another hydrophobic polymer, lignin, and had intermediate adsorptive abilities which were not significantly different from one another. These preparations modelled lignified cell walls which occur in the diet, for example in wheat bran. Our results indicate that suberized and lignified cell walls may be important in protecting against colorectal cancer.