Background: Surveillance for CNS metastasis (mets) is not routinely performed in pts with clinically localized CM. Improved understanding of the incidence, timing and risk factors for the development of CNS metastasis in these pts may inform surveillance strategies. Methods: Under an IRB-approved protocol, demographics, tumor characteristics, and clinical events were collected for pts diagnosed from 1998 to 2019 with AJCC 8th edition stage I or II CM at MD Anderson Cancer Center. Dates of initial diagnosis, regional, distant non-CNS, and CNS mets were recorded. Symptoms and the extent of disease (brain, LMD, both) were recorded for pts with CNS mets. Cumulative incidence of distant mets (CNS and non-CNS) was determined using the competing risks method, including death; pts without CNS mets and alive at last follow-up were censored. Differences in cumulative incidence between groups were assessed using Gray’s test. Associations between measures of interest and cumulative incidence were determined using proportional subdistribution hazards regression models. All statistical tests used a significance level of 5%. Results: 5,179 Stage I-II CM pts were identified. At a median follow up of 82 (0.0-268.8) months, 703 (13.6%) pts were diagnosed with distant mets, including 355 (6.9%) with CNS mets. Cumulative incidence of CNS mets was 0%, 2%, and 5% at 1, 2, and 5 years, respectively. Among pts with distant mets, the first site of distant mets was CNS only for 29 (4%), non-CNS only for 557 (79%), and both for 116 (17%) pts. At initial diagnosis of CNS mets, 195 (55%) pts were asymptomatic, and 46 (13%) had no active extracranial disease. Median time to any distant met was longer for pts who were diagnosed with CNS mets [40.0 (1.9-238.0) months] vs pts diagnosed with non-CNS mets only [31.4 (1.1-185.7) months, p < 0.001]. On multivariable analysis, risk of CNS mets was significantly associated with primary tumor location of scalp [Hazard Ratio (HR) 3.4, 95% Confidence interval (CI) 1.9-5.9], head/neck (HR 3.3, 95% CI 2.0-5.3), or trunk (HR 2.3, 95% CI 1.5-3.5) (vs upper extremity); acral lentiginous melanoma subtype (HR 2.0, 95% CI 1.2-3.6) (vs superficial spreading); increased T category (T2 HR 1.5, 95% CI 1.1-2.2; T3 HR 1.9, 95% CI 1.2-3.0; T4 HR 2.1, 95% CI 1.1-3.8; vs T1), Clark level (CL) (CL4 HR 2.1, 95% CI 1.2-3.7 vs CL2), and mitotic rate (MR) (MR 5-9/mm² HR 2.1, 95% CI 1.5-3.0; MR > 9/mm² HR 2.0, 95% CI 1.3-3.0; vs MR 0-4/mm²). While high ( > 9/mm²) MR was associated with increased risk of CNS and non-CNS mets, intermediate (5-9/mm²) was associated with CNS mets only. Conclusions: Primary tumor location, tumor thickness, and MR were strongly associated with risk of CNS mets. MR rate was more strongly associated with risk of CNS than non-CNS mets. Validation in independent cohorts may provide evidence to support CNS surveillance strategies in select pts with stage I-II CM who are deemed high risk for CNS mets.