The manifestations of mycosis fungoides in its early stage may mimic clinically and histologically those of many benign inflammatory dermatoses. Therefore, the diagnosis of mycosis fungoides remains a major challenge for dermatologists and dermatopathologists. For many years, it has been proposed that atypical lymphocytes within the epidermis constitute one of the diagnostic features in mycosis fungoides. Presence of dermal atypical lymphocytes remains controversial as a diagnostic criterion. We reassessed the feasibility of applying lymphocytic atypia within epidermis and dermis as diagnostic criteria discriminating between mycosis fungoides and spongiotic dermatitis. Thirty cases of mycosis fungoides and 30 cases of spongiotic dermatitis were retrieved from archival hematoxylin and eosin-stained histologic sections. Punch biopsy sections were examined by light microscopy; epidermal and dermal lymphocytes were photographed at 1000x (oil immersion). A total of 92 ektachrome slides (35 mM) were developed, coded, and ordered randomly. For each slide, cells were interpreted as typical or atypical lymphocytes by seven pathologists. Atypical epidermal lymphocytes were judged to be present in 9 +/- 2 out of 16 (56%) cases of mycosis fungoides photographed as compared with 8 +/- 3 out of 16 (50%) in spongiotic dermatitis. Dermal lymphocytic atypia was thought to be present in 14 +/- 6 out of 30 (47%) patients with mycosis fungoides. Thirteen +/- 6 out of 30 (43%) patients with non-mycosis fungoides also displayed dermal lymphocytic atypia. No statistical significance was observed in these comparisons (t test, P >.05). Furthermore, atypia of lymphocytes was deemed to be present in 41, 38, 59, 70, 23, 47, and 40 out of 92 slides examined by the investigators, suggesting that observer variation is a very significant factor in our present study. We conclude that it is not possible to distinguish mycosis fungoides from spongiotic dermatitis merely based on lymphocytic atypia within epidermis or dermis.