Convincing results from randomized controlled trials (RCTs) have led to increasing use of immune checkpoint inhibitors (ICI) as part of standard therapies in real-world (RW) scenarios. However, RW patients differ clinically from RCT populations and might have reduced long-term survival. Currently, only sparse data on 3-5-year survival rate for RW patients with advanced non-small cell lung cancer (NSCLC) treated with ICI exist. A multicenter study was performed including 729 patients with advanced NSCLC receiving monotherapy with ICI (retrospective data (n = 566) and prospective data (n = 163)). Detailed baseline clinical characteristics, programmed death-ligand 1 (PD-L1) tumor proportion score (TPS), and baseline haematological count were registered. Kaplan–Meier estimates and log-rank test were used for survival analyses, Cox regression for determination of prognostic factors. Median time of follow-up (FU) was 48.7 months (IQR 37.2–54.3). Median overall survival (OS) in first line treatment was 20.4 months (IQR 8.5–45.0) compared to 11.4 months (IQR 4.6–27.1) in ≥2nd line (HR 1.48, 95% CI 1.25–1.75). Estimated probability of OS was 30% at 3 years, 23% at 4 years, and 13% at 5 years in first line compared to 17, 13, and 11% in ≥2nd line, respectively. For those with performance status (PS) 2, the 2-year OS rate was 32% (95% CI 0.22–0.43) compared to 5% (95% CI 0.01–0.15) in patients with PD-L1 ≥ 50% versus <50%, respectively. Compared to RCTs, long-term OS and PFS rates are lower in real-world patients treated with ICI in first line but much improved compared to historic rates on chemotherapy. A promising flattening of both the OS and progression free survival curves illustrates that also a subset of real-world patients obtain long-term remission. Patients with PS 2 and PD-L1 ≥ 50% may obtain clinically meaningful 2-year PFS and OS rates.