Polycyclic aromatic hydrocarbons (PAH) are pollutants frequently found in soils, particularly in urban areas. From polluted soils, the PAH can be taken up into crops and the consumption of these crops can result in a human health risk. We estimated the bioconcentration factors (BCF) in the edible plant tissues for PAH using crop-specific models for leafy vegetables, root vegetables, potatoes and tree fruits. The estimates were compared with results from the empirical regression of Travis and Arms (T&A) for above-ground vegetation. The comparison shows that the use of crop-specific models resulted in lower BCF values of pollutant concentrations in fruits, potatoes and leafy vegetables, particularly for the heavier PAH (M>220 g mol(-1)). However, the crop-specific models yielded higher BCF values for root vegetables (carrot) and leafy vegetables if the attached soil particles (1%) were considered. Consequently, the average daily intake of benzo(a)pyrene (BaP) by an adult Czech through fruits and vegetables was estimated with the crop-specific models to be 190 ng BaP per person, and 460 ng BaP per person with the T&A regression, for a soil concentration of 1 mg BaP kg(-1) soil (wet wt.). A virtually safe oral dose of BaP, as a marker of the carcinogenic PAH, was suggested by a European expert commission to be below 4.2 to 35 ng per person and day. Using these figures, an acceptable soil concentration of BaP was estimated for the purpose of crop production to be below 0.02 or 0.18 mg kg(-1) (wet wt.) with the crop-specific models, and below 0.01 or 0.08 mg kg(-1) (wet wt.) with T&A. The results demonstrate clearly the advantage of the crop-specific exposure assessment: it can be adapted to different food baskets and allows more effective risk assessment and management of soil pollution.