Background: Recent data suggest that patients with stage III melanoma are at high enough risk for developing CNS metastases to consider routine surveillance neuroimaging (Journal of Clinical Oncology; PMID: 31990608). Given that a subset of stage II patients have a worse prognosis than stage III patients, we investigated the risk of developing brain metastases in stage II disease and compared it to the risk in stage III disease. Methods: We studied a cohort of prospectively enrolled melanoma patients who had protocol driven follow-up at New York University (NYU) Langone Health. We investigated both the incidence and time to development of CNS metastases, and explored whether the frequency of CNS metastases as a first isolated site of distant disease varies among the different stages. Results: The study cohort included a total of 1,102 patients (stage II: n = 619 with median follow-up 56.5 months; stage III: n = 483 with median follow-up 40.9 months). 85/619 (14%) stage II and 91/483 (19%) stage III patients developed CNS metastases (p = 0.03). The estimated 5-year cumulative incidence was 9% in stage IIA, 14% in stage IIB, and 29% in stage IIC patients (p = 0.0001). It was 10% in stage IIIA, 32% in stage IIIB, 23% in stage IIIC, and 49% in stage IIID (p = 0.0001). The CNS was the site of first metastasis for 32/154 (21%) stage II patients who developed distant disease compared to 28/214 (13%) stage III patients (p = 0.06). Conclusions: A subset of stage II patients are at an elevated risk for developing CNS metastases within 5 years of their initial diagnosis, which is comparable to that seen in stage III patients. The frequency of the CNS as a first site of metastasis in stage II melanoma suggests a propensity for brain tropism that cannot only be explained by a generalized pro-metastatic phenotype. Surveillance strategies that incorporate serial neuroimaging should be considered for these individuals.