Serum samples from 274 patients allergic to one or more of three pollens (birch, grass, mugwort), from 36 patients allergic to cat and/or Dermatophagoides pteronyssinus but not to pollen and from 55 non-allergic controls, as well as 20 cord blood samples, were examined for specific IgE to six 'pollen-associated' food allergens by using a new sensitive assay (CAP). A questionnaire asking for reactions to food was also sent to all patients. In the pollen group, 111 patients (47%) were positive (> or = 0.71 kU/l) for a food allergen (392 positive tests). Of these, 92 were sensitive to apple, 68 to potato, 64 to carrot, 63 to celery, 61 to peach and 44 to melon. In the non-allergic group, no IgE to any of the food allergens tested was found, whereas in the group allergic to non-pollen allergens, only one individual had such an IgE. The CAP assay was found to be more sensitive than RAST for the allergens studied. A history of clinical reactions (oral symptoms in 67, rhinoconjunctivitis in 65, asthma in 42 and urticaria in 39) to the corresponding food allergen was reported mainly by patients with positive CAP. In conclusion, we found a high prevalence of IgE to some food allergens in patients allergic to pollen and the absence of such antibodies in the control groups. The new in vitro assay, being more sensitive than previous ones, indicated a high prevalence of food specific IgE in pollen allergic patients, which in many cases did not correspond to clinical symptoms of food allergy.