Background: Lung cancer is the leading cause of cancer death in the United States, with > 85% classified as non-small cell lung cancer (NSCLC). Diabetes mellitus (DM) is a common comorbidity in patients with NSCLC. While surgery is the standard of care for early-stage NSCLC, patients who have DM with end organ damage are considered medically inoperable according to treatment guidelines and whether this influences NSCLC treatment and outcomes is unclear. This study aimed to investigate treatment patterns and outcomes among patients with early-stage NSCLC and DM. Methods: Using the Surveillance, Epidemiology, and End Results database linked to Medicare (2000-2016), we identified patients ≥65 years old with Stage I-IIIA NSCLC treated with lobectomy, limited resection (wedge resection and segmentectomy), or no surgery. DM and complications at the time of NSCLC diagnosis were ascertained through published claims-based algorithms. Patients were categorized as having no DM, DM without severe complications (DM_-c), or DM with ≥1 severe complication (i.e., end-organ damage, DM_+c). We used multinomial logistic regression to assess if DM was associated with treatment. Association of DM with overall survival (OS) and lung cancer-specific survival (LCSS) was analyzed with Cox regression stratified by treatment type. These analyses controlled for demographics, comorbidities, and NSCLC histology and stage. Results: Of 60,300 patients analyzed, 45,270 (75%) had no DM, 6,873 (12%) had DM_-c and 7,887 (13%) had DM_+c. More patients with DM_+c (N = 4,508[57%]), did not receive surgery vs. patients with DM_-c (N = 3,771[55%]) and without DM (N = 23,289[51%]). DM was associated with lower odds of receiving lobectomy vs. no surgery in adjusted analysis (odds ratio [OR]: 0.88; 95% confidence interval [CI]: 0.83-0.93 for DM_-c, and OR: 0.91; 95% CI: 0.86-0.97 for DM_+c vs. no DM), but not for limited resection vs. no surgery (OR: 0.92; 95% CI: 0.83-1.02 for DM_-c and OR: 0.91; 95% CI: 0.92-1.11 for DM_+c vs. no DM). Cox regression showed that in patients with lobectomy and limited resection, compared to no DM, DM_+c was associated with worse OS (hazard ratio [HR]: 1.21; 95% CI: 1.15 to 1.27 [lobectomy]; HR: 1.17; 95% CI: 1.07-1.28 [limited resection]), but not LCSS (HR: 1.06; 95% CI: 0.99-1.14 [lobectomy]; HR: 1.02; 95% CI: 0.90-1.17 [limited resection]). Among patients who received no surgery, DM_+c patients had both worse LCSS (HR: 1.05; 95% CI: 1.00-1.09 and OS (HR: 1.12; 95% CI: 1.08-1.16) vs. no DM. DM_-c was not associated with worse LCSS or OS for all treatment categories. Conclusions: Patients with Stage I-IIIA NSCLC and DM_+c were less likely to undergo surgery and had worse OS but not LCSS if they underwent full or limited resection, while they had worse OS and LCSS if they did not have surgery. These findings suggest that patients with DM with end-organ damage benefit from more aggressive NSCLC treatment, but research is needed to determine optimal treatments in these patients.