Maspin is a serine protease inhibitor that is thought of as a tumor suppressor because of observations that loss of maspin expression in breast, prostate, and oral cancer is associated with poor prognosis. In addition, maspin may function as an inhibitor of angiogenesis. However, it has been correlated with malignant behavior in pancreatic and ovarian cancer. The role of maspin in malignant melanoma (MM) has not yet been systematically examined. We aimed to examine the immunohistochemical expression of maspin and several proangiogenic factors (vascular endothelial growth factor, transforming growth factor-beta, alphaVbeta3 integrin, cyclooxygenase-2, and CD44) in MM and correlate each to angiogenesis, tumor thickness, and outcome. In all, 77 formalin-fixed, paraffin-embedded MM samples were immunostained for maspin and other proangiogenic factors (vascular endothelial growth factor, transforming growth factor-beta, alphaVbeta3 integrin, cyclooxygenase-2, and CD44) and were correlated with angiogenesis as mean microvessel density. Three normal-appearing skin samples and 10 nevi were also immunostained for maspin. Breslow thickness, Clark level, clinical stage, and follow-up information were obtained for outcome analysis. Immunohistochemical analysis revealed strong nuclear melanocytic maspin expression in all 10 nevi (half of which were dysplastic) but none in melanocytes from 3 normal-appearing skin samples. Strong nuclear maspin staining was demonstrated in 78% of radial phase melanoma and 46% of vertical growth phase melanoma. In addition, there was a significant inverse relationship between maspin and microvessel density (P = .018) and tumor thickness greater than 0.76 mm (P = .007), indicating that maspin is expressed in thinner tumors with less angiogenesis. Conversely, vascular endothelial growth factor expression, Clark level, and Breslow thickness all significantly correlated with microvessel density (P = .047, P = .027, and P = .011, respectively). Cyclooxygenase-2 expression significantly correlated with thicker tumors (P = .006) but not with angiogenesis (P = .714). In addition, Clark level, Breslow thickness, and stage were all significant predictors of overall survival (P < .001, P = .005, and P < .001, respectively). This study represents a single institution. These results demonstrate maspin expression in nevi and radial growth phase melanoma, but this expression seems to be lost in the transition from radial growth phase to vertical growth phase melanoma. In addition, maspin is correlated with decreased angiogenesis and tumor thickness less than 0.76 mm in MM. These results indicate maspin may function as a tumor suppressor in MM.