We evaluated a new method of computer-based scanning of skin prick test wheal areas. To test the calibration of the program, we scanned five constructed circles of exactly defined areas between 5 and 255 mm2. One of these circles was scanned in different degrees of contrast (modes), the middle mode being used in the other experiments. We also investigated the inter- and intraoperator variation. Finally, results obtained by measuring diameters of wheals and by the new method of area determination were compared for 82 histamine and 75 egg wheals. The areas found agreed well with the real areas (P < 0.01) (mean 97.4-100.8%), except for the smallest wheal (5mm2) (NS). Areas obtained in the "middle position" closely resembled the real area. The intraoperator coefficient of variation (c.v.) was 1.4% (0.3-4.3), the day-to-day c.v. was 1.9% (0.2-5.3), and the interoperator c.v. was 2.3% (0.7-5.7), with a higher c.v. for small areas. Histamine and egg wheals were significantly larger with the diameter method (126% and 129%, respectively) than with the area method. The difference between the methods was most pronounced for small wheal areas. The c.v. of the scanning method was significantly lower than that of the diameter method. The new program was precise and is recommended for registration of skin test reactions in scientific trials. Although the scanner works well and has high precision, the major problem in skin testing seems still to be the reproducibility of the skin test technique employed.