Although osimertinib (osi) has excellent intracranial activity in EGFR-mutant metastatic non-small cell lung cancer (NSCLC), there is no consensus regarding whether to continue osi for central nervous system (CNS) control with second-line chemotherapy (chemo) at the time of systemic progression. We aimed to compare CNS outcomes after first-line osi in patients receiving second-line chemo with or without continuation of osi. We retrospectively reviewed patients with EGFR-mutant NSCLC with brain metastases (BrM) at the time of initiating first-line osi who experienced progression and started second-line chemo. Cumulative incidence of local and distant CNS progression, and extracranial (EC) progression was calculated from time of second-line chemo initiation with death as a competing risk. Overall survival (OS) was analyzed using Kaplan-Meier. We included 52 patients with a median follow up of 9.6 months (range 0.4-36.4). Median OS and CNS progression-free survival (PFS) from the time of starting second-line chemo was 12.5 months (95% CI 8.1-16.9), and 5.3 months (95% CI 3.35-7.26), respectively. The 1-year cumulative incidence of local, distant CNS progression, any CNS progression, and EC progression was 14.4% (95% CI 4.5-24.2), 42.8% (95% CI 22.8-56.8), 42.8% (95% CI 22.8-56.8) and 66.8% (95% CI 53.5-80.2), respectively. After progression on first-line osi, 25 (48.1%) and 27 patients (51.9%) continued and discontinued osi, respectively. Patients who continued osi had significantly higher BrM burden than those who did not, with 17 (68%), 3 (12%), and 5 (20%) versus 26 (96%), 0, and 1 (3.7%) patient having <10 or >11 parenchymal brain lesions, or leptomeningeal disease (LMD) at the time of second line therapy, respectively (p<0.01). In those who continued osi vs those who did not, median OS (10.8 vs 12.5 months; p = 0.37), median intracranial PFS (5.3 vs 4.8 months; p = 0.99), 1-year cumulative incidence of local (8.4% versus 20 % p = 0.26), and 1-year distant CNS progression (24.8% vs 60%; p = 0.08) was not significantly different. CNS complications such as symptomatic, hospitalizations, and steroid initiation for CNS disease, and progression of LMD were not significantly different between the two groups. Eventually, 10 patients underwent salvage RT post first-line osi and median time to salvage RT was 7.8 months (range 2-9.4). Of patients who underwent salvage RT, 2 patients (20%) had continued osi with second-line chemo. Twelve patients (44.4%) who did not continue osi eventually re-started osi for progressive disease. Patients who continued osi had significantly higher BrM tumor burden. Despite these patients being at higher risk for CNS progression, time to CNS progression and incidence of CNS complications were not significantly different in the two cohorts. Patients who discontinued osi were more likely to undergo salvage RT. Continuation of osi may allow patients to defer salvage RT.