OBJECTIVES: Guideline recommendations for first-line non-small cell lung cancer (NSCLC) therapy are partially based upon physician-reported performance status (PS). Lack of PS in claims data limits evaluation of guideline recommended initiation (GRI) of therapy. This study adapted a previously validated claims-based algorithm for PS to assess GRI. METHODS:  Patients with metastatic NSCLC and Medicare coverage were identified from Humana medical claims and 2013-2014 treatment preauthorizations for infusion therapy. Index was the first infusion date. PS was imputed from a classification model using procedure, diagnosis and durable medical equipment codes one year pre-index. The model predicted poor (PS=3-4) vs good/moderate PS (PS=0-2). Moderate PS (PS=2) was estimated using the top quintile of good/moderate PS. Six-month post-index mortality and hospice status validated PS cut-points. GRI was ≥1 cycle of National Comprehensive Cancer Network recommended first-line therapy based on age and PS or targeted therapies regardless of age and PS. Significance was from Chi-square and ANOVA. RESULTS: For 1,353 eligible patients with metastatic NSCLC, mean (SD) age was 71.4 (6.6), 45.9% were female. After applying the PS algorithm, 75.8% had good, 18.8% had moderate and 5.4% had poor PS. Mean Deyo-Charlson Comorbidity score was 8.0 (3.3) with no difference by PS. Six-month mortality was 23.0% for good, 27.2% for moderate and 37.0% for poor PS (p=0.020). Hospice was initiated by 8.8% of patients with good PS, 13.0% with moderate PS and 19.2% with poor PS (p=0.008). GRI was observed in 87.7% of patients with good PS, 46.1% with moderate PS, and 4.1% with poor PS (p<0.001). CONCLUSIONS: Determining PS from claims data allows assessment of GRI of first-line therapy for metastatic NSCLC. Increased rates of mortality and hospice with worsening PS suggests that cut-points used to distinguish poor, moderate and good levels of PS may offer discrimination of PS categories and determination of GRI from claims data.