In 60 healthy adult volunteers and 58 patients with gastrointestinal disease a test system (Multitest) consisting of a plastic disposable multiple-puncture device capable of simultaneously applying seven delayed-type hypersensitivity antigens and a glycerin/saline diluent (negative control) was assessed. The Multitest device was applied on both the inner side of the forearm and on the back for assessment of cell-mediated immunity (CMI). The antigens used were two toxoids, tetanus and diphtheria, three bacterial antigens, Streptococcus, tuberculin and Proteus and two fungal antigens, Candida and Trichophyton. A scoring system based on both number and size of positive response revealed a median "score" on arm and back of 19 mm and 14 mm respectively, in the healthy volunteers and a median "score" of 12 mm and 8 mm respectively in patients with gastrointestinal disease. In both groups a significant difference was found between back and arm (P less than 0.01). The coefficient of determination (r2) shows that only 64% of the variability in scores on the back is explained by the regression line. Therefore, scores obtained from tests on the back cannot be interpreted with reference to normal values originating from tests applied to the inner side of the forearm.