Background: Allergy to tree nuts, like hazelnuts, ranks among the most frequently observed food allergies. These allergies can start at early childhood and are, in contrast to other food allergies, not always outgrown by the patient. Tree nut allergy is frequently associated with severe reactions. Diagnosis partially relies on in vivo testing by means of a skin prick test (SPT) using commercially available SPT reagents. Methods: Protein and allergen composition of nine commercial SPT solutions was evaluated using standard protein detection methods and specific immunoassays for measurement of five individual allergens. Diagnostic performance was assessed by SPT in 30 hazelnut-allergic subjects, of which 15 were provocation proven. Results: Protein concentrations ranged from 0.2-14 mg/ml. SDS-PAGE/silver staining revealed clear differences in protein composition. The major allergen Cor a 1 was present in all extracts but concentrations differed up to a factor 50. An allergen associated with severe symptoms, Cor a 8 (lipid transfer protein), was not detected on immunoblot in three products, and concentrations varied by more than a factor 100 as was shown by RAST inhibition. Similar observations were made for profilin, thaumatin-like protein and a not fully characterized 38-kD allergen. Ratios of individual allergens were variable among the nine extracts. SPT showed significant difference, and 6/30 patients displayed false-negative results using 3/9 products. Conclusion: Variability in the composition of products for the diagnosis of hazelnut allergy is extreme. Sometimes, allergens implicated in severe anaphylaxis are not detected by immunoblotting. These shortcomings in standardisation and quality control can potentially cause a false-negative diagnosis in subjects at risk of severe reactions to hazelnuts. Copyright © 2003 S. Karger AG, Basel.