Background: A 31-gene expression profile (GEP) that provides a binary outcome of low or high risk of distant metastasis (DM) within 5 years of diagnosis has been validated as an independent predictor of distant metastasis free (DMFS) and overall survival (OS). We present analysis of DM prognostication in a previously unreported cohort of cutaneous melanoma (CM) patients. Methods: 334 CM primary tumors from 12 centers were analyzed as part of an IRB-approved archival tissue study. Quantitative RT-PCR and predictive modeling were performed as previously described (Gerami, JAAD, 2015) and classified tumors as low risk Class 1 or high risk Class 2. Results for Cox regression survival analysis are reported. Results: Cox univariate regression analysis of the cohort indicated that Breslow’s thickness (BT), ulceration (U), mitotic rate (MR), sentinel lymph node (SLN) status and GEP class were significant predictors of DM risk (p < 0.008 for all; median BT = 1.5mm, median follow up = 5.3 years). 156 (47%) cases with documented BT, U, MR, SLN status and GEP class were included in multivariate analyses. A documented distant metastatic event was reported for 46 of 156 (29%) cases. Multivariate analysis of all factors resulted in a significant association of SLN status (p = 0.002) and GEP (p = 0.031) with DM (table). 13 of 83 (16%) SLN negative cases had a DM event, and 10 (77%) of these had a Class 2 GEP result. Accuracy of DM prediction by GEP showed 76% sensitivity, 55% specificity, 42% positive and 85% negative predictive values compared to 72%, 64%, 45% and 84%, respectively, for SLN. Conclusions: GEP offers prognostic information that complements conventional staging in order to better identify CM patients at risk for distant metastasis. Results achieved in this multicenter study parallel previously reported performance of the GEP. Prospective evaluation of the utility of closer surveillance for detection of DM, and the potential benefits of adjuvant treatment, for Class 2 patients is warranted.