Background: Dermoscopy alone is not sufficient to detect all early melanomas. Total body photos reveal growth of melanomas but also reveal growth of melanocytic nevi. Objective: We set out to determine whether a simplified algorithm on the basis of nonuniform dermoscopic features combined with growth noted from baseline total body photos targeted the early melanocytic neoplasms most likely to be malignant. Methods: Lesions removed during follow-up of patients with total body photographs were reviewed and 169 melanocytic lesions were identified for which both gross clinical and dermoscopic photos were available. The images were evaluated separately by 3 academic dermatologists, without knowledge of the given pathologic diagnosis, for uniformity (consistent gradient of features from center to edge) and change (specifically, that which could indicate melanoma growth in normal skin or within a nevus). Results: Using a minimum of 2 out of 3 agreement for uniformity and change, 12 of 16 melanomas (including all 5 superficially invasive tumors) were graded as nonuniform and changed. The 4 melanomas not included in this category were in situ. The predicted odds of melanoma for lesions scored as both nonuniform and changed was 4.06 (P > .0195). If 3 out of 3 agreement was used, the odds ratio increased to 6 (P > .0010). Conclusion: An algorithm on the basis of dermoscopic nonuniformity and change suggestive of growth as determined by total body photography segregates melanocytic neoplasms most likely to be malignant. (J Am Acad Dermatol 2003;48:663-71.)