We compared two techniques for detecting delayed-type hypersensitivity (DTH) skin responses in 359 patients infected with human immunodeficiency virus (HIV) (mean CD4+ lymphocyte count, 387/microL). DTH responses were assessed with use of two antigenic panels administered simultaneously: tuberculin purified protein derivative (PPD) plus three control antigens (Candida albicans, mumps antigen, and tetanus toxoid) administered by the Mantoux method and by a multiple-puncture device delivering seven antigens percutaneously (MULTITEST CMI; Institut Mérieux, Lyon, France). Eighty-three patients (23%) were anergic, 216 (60%) reacted to both panels, 55 (15%) did not react to MULTITEST CMI but did react to the antigens administered by Mantoux method, and only five (1%) reacted to MULTITEST CMI without reacting to antigens administered by the Mantoux method (P < .001, McNemar's test). Each of the three possible combinations of PPD plus two control antigens administered by the Mantoux method were also superior to MULTITEST CMI for classifying patients as nonanergic (P < .001, McNemar's test). We conclude that the application of antigens by the Mantoux method is more efficient than MULTITEST CMI for detecting DTH skin responses in HIV-infected patients.