Background: National guidelines recommend biomarker testing in mNSCLC, and targeted therapy is associated with improved outcomes. The aim of this study was to understand the real-world biomarker testing and treatment patterns in the community setting. Methods: This was a retrospective study of adult patients diagnosed with de novo mNSCLC between 01-Jan-2016 and 30-Sep-2019, with follow-up through 31-Dec-2019 using The US Oncology Network structured electronic health records data. Patients who received systemic treatment for mNSCLC were included. Results: A total of 3213 patients were identified with median age 68 years (24, 90+); 52.7% were male and 10% were current smokers. ECOG score was 0-1 in 55.2%; 60% had adenocarcinoma, 16% had squamous cell carcinoma, and the rest had other/unknown histology. Since most of the biomarker-guided therapies were approved after 2016, testing patterns are described for 2017-2019 (n=2257). Overall, 23.6% were not tested for any biomarker (PD-L1 or driver mutation [DM]) at any time during the study period, and only 49% had a biomarker test result prior to 1L treatment. We observed similar patterns when assessing DM specifically; 35.8% were never tested for DM, and only 39.3% had a DM test result prior to 1L treatment. As an example, out of 42 ALK+ patients in this study population, only 5 had test results prior to 1L treatment and only 3 received an ALK inhibitor as their 1L treatment (Table). Similar patterns were observed for the other biomarkers. Conclusions: Despite availability of promising biomarker-based therapies, the lack of adequate testing in the community oncology setting means that not all eligible patients are receiving the most effective therapies upfront. Nearly 61% of patients had no DM test reported before 1L treatment in this mNSCLC cohort (all histologies), and some were determined to be DM positive at a later time, highlighting a missed opportunity to employ the most effective biomarker-directed front-line treatment. Next steps in this study will include assessing patterns by histology. Structured data, which are recorded for clinical management, might have gaps; future research with chart reviews could provide a more comprehensive assessment.