In order to establish the most efficient thresholds for serum allergen-specific IgE measured by the Pharmacia CAP System, a "second-generation" in vitro method, we evaluated results from 89 subjects with suspected inhalant allergies, using receiver-operating characteristic (ROC) curve analysis. Sera samples were analyzed by the Pharmacia CAP System for specific IgE, the choice of allergens to be tested being based upon the symptoms and clinical history of each patient. Results were analyzed by ROC analysis for the five most representative allergens, cat dander (e1), Dermatophagoides pteronyssinus (d1), Lolium perenne (g5), wall pellitory (W19), and wormwood (w5). The areas under the ROC curves were found to be satisfactory, ranging from 0.931 (e1) to 0.974 (g5) when we excluded w5, which had a significantly smaller area (0.81). To establish the most efficient threshold for each allergen, we calculated the clinical sensitivity, specificity, efficiency, and negative and positive predictive values. The thresholds giving a higher diagnostic efficiency were as follows: 0.40 kUa/l for d1, 0.55 kUa/l for g5, 0.50 for e1, 0.65 kUa/l for w19, and 1.00 kUa/l for w5. It is concluded that quantitative reporting of specific IgE measurement has numerous advantages, but the choice of the positive threshold seems to be a prerequisite for obtaining the optimal clinical efficiency. It is also suggested that a specific threshold should be adopted for each allergen.