OBJECTIVE To compare the deposition of IgA and C3 in the skin of patients with active dermatitis herpetiformis relative to the sites of disease. METHODS In the phase 1 study, skin biopsy specimens were obtained from erythematous perilesional skin, nonerythematous perilesional skin, and never-involved skin. In the phase 2 study, specimens from the nonerythematous perilesional and uninvolved skin from the same anatomic region were sampled. METHODS The Dermatology Clinic at the University of Utah Health Sciences Center, Salt Lake City. METHODS Patients with known dermatitis herpetiformis: 19 patients in the phase 1 study and 15 patients in the phase 2 study. Suppressive medications were stopped for 48 to 72 hours after biopsy specimens were obtained. All patients had active disease at the time that biopsy specimens were taken. METHODS The intensity of IgA and C3 immunofluorescent staining in 6 sections from each skin biopsy specimen was graded by using a semiquantitative scale (0 to 3+) in a blinded fashion by a single observer. RESULTS Deposition of IgA was more intense in noninflamed perilesional skin in 11 of 19 patients compared with that in erythematous skin (P < .05). Erythematous skin was negative for IgA in 16% (3/19) of the specimens. Noninflamed perilesional skin showed more intense IgA deposition in 18 of 19 specimens compared with that in never-involved skin (P < .01); C3 was more intense in erythematous skin (P < .01). In the phase 2 study, skin from the same anatomic region revealed greater deposition of IgA near lesions in 12 of 15 patients (P < .001). CONCLUSIONS In patients with dermatitis herpetiformis, IgA is not uniformly distributed throughout the skin, and IgA is present in greater amounts near active lesions. The preferred biopsy site for the diagnosis of dermatitis herpetiformis is normal-appearing skin that is adjacent to an active lesion.