Stimulating the skin with intensities close to the sensory threshold causes erroneous localization of the site of stimulation. Previous studies using manual methods for applying faint tactile stimuli have shown that localization errors obey a somatotopic principle in which tactile stimuli are preferentially mislocalized to sites adjacent to the stimulated skin region. However, manual testing of mislocalization is time consuming and only partially objective because results depend on the skills of the tester. To improve the testing procedure, an automated apparatus was developed. The procedure adjusted stimulus intensity adaptively during testing to remain near the individual subject's sensory threshold, so that mislocalizations occurred often enough to assess somatotopic organization. The new method was applied to 12 healthy subjects. In each subject, the five digits of the right hand were stimulated singly in random order. Localization errors were distributed preferentially to fingers close to the stimulated finger rather than to distant fingers. The profile of mislocalization differed significantly from that expected on the basis of response bias or guessing behavior. The present results replicate previous findings obtained for manual testing with improved sensitivity and indicate that the new technique is a useful tool for the study of somatosensory processing on a perceptual level.