To assess the diagnostic accuracy of margin evaluation of melanocytic lesions using en face frozen sections compared with standard paraffin-embedded sections, we studied 2 sets of lesions in which en face frozen sections were used for analysis of surgical margins (13 from malignant melanomas [MMs] and 10 from nonmelanocytic lesions [NMLs]). Routine permanent sections were cut after routine processing. The slides were mixed and coded randomly. Fifteen dermatopathologists examined the cases separately. Margin status was categorized as positive, negative, or indeterminate. Kappa statistics were calculated per dermatopathologist and per case. One case from each group was excluded because epidermis was not available in the routine sections. Of 330 evaluations (22 cases, 15 dermatopathologists), there were 132 diagnostic discrepancies (40.0%): 66 each for MM and NML (mean per case for both diagnoses, 6). In 9 instances (6.8%), the change was from positive (frozen) to negative (permanent) and in 43 (32.6%), from negative (frozen) to positive (permanent). There was poor agreement between frozen and permanent sections (kappa range per dermatopathologist, -0.1282 to 0.6615). If permanent histology is considered the "gold standard" for histologic evaluation, en face frozen sections are not suitable for accurate surgical margin assessment of melanocytic lesions.