Although most examples of cutaneous malignant melanoma are easily recognized by their clinical appearances, in some cases this serious neoplasm may clinically simulate other less serious forms of skin cancer or benign processes. This study was undertaken to assess both the sensitivity of clinical diagnosis of cutaneous malignant melanoma and the efficacy of biopsies of clinically unsuspected melanomas in yielding specimens on which complete and accurate histologic assessments could be made. A retrospective analysis of 1784 cases of histologically proven melanomas diagnosed between 1985 and 1990 was performed in search of lesions not clinically suspected. Biopsy techniques used to sample these lesions were subjected to critique of their efficacy in yielding specimens that could be accurately diagnosed and completely assessed histologically. Of 1784 histologically proven primary cutaneous melanomas, 583 were not clinically suspected, yielding a sensitivity of 67%. Clinical diagnosis included nevi (33%), no diagnosis (17%), multiple diagnoses (13%), basal cell carcinoma (12%), keratosis (9%), and lentigo (9%) among others. The biopsy methods used to sample these lesions were shave (56%), excisional (24%), punch (11%), curettage (2%), and undetermined (6%). Eighty-six percent of shave biopsies could be accurately assessed while only 32% of punches and no curettages provided sufficient material for both definitive and complete evaluation of melanomas. Eighteen percent of specimens histologically reviewed were considered inadequate for complete evaluation. In 34%, the actual diagnosis of melanoma was uncertain because of inability to assess diagnostic features as a consequence of the biopsy technique. Melanoma may be unsuspected clinically in a significant number of cases and may be mistaken for less serious cutaneous neoplasms.(ABSTRACT TRUNCATED AT 250 WORDS)