Background: Cancer patients with pre-existing autoimmune diseases (AID) have been traditionally excluded from clinical trials of immune checkpoint inhibitors (ICI) due to concerns for immune activation leading to toxicity. As indications for ICI expand, there is a need for robust data on safety and efficacy of ICI in cancer patients with AID. Existing studies examined a heterogenous group of cancers and do not include a comparison to cancer patients without pre-existing AID. Therefore, data are lacking concerning how safety and efficacy of ICI in this group of patients may differ from the general population and whether the results are generalizable to non-small cell lung cancer (NSCLC). Methods: We searched for studies consisting of NSCLC, AID, ICI, treatment response, and adverse events using database-controlled vocabulary terms. We systematically searched Medline (PubMed), EMBASE, Scopus, CINAHL, and Web of Science. We selected studies that included NSCLC and excluded abstracts, case reports, review articles, and articles lacking outcomes or populations of interest. Three authors (WA, CL, NS) independently reviewed all abstracts to determine study eligibility and quality. Study quality was assessed using the Newcastle-Ottawa Scale. Study data were pooled using random-effects meta-analysis. Results: Data were extracted from 24 cohort studies, consisting of 10,924 cancer patients, of which 4353 were NSCLC patients. Studies examined a broad spectrum of AID consisting of 1157 patients, of which 291 were NSCLC patients. Pooled analysis revealed an AID flare incidence of 36% (95%CI 27%-46%) in all cancers and 23% (95%CI 9%-40%) in NSCLC. Tests of subgroup difference revealed no significant difference in AID flares by organ system in all cancers and NSCLC (p = 0.602 and p = 0.19, respectively). Pre-existing AID was associated with a higher risk of de novo iRAE in all cancer patients (RR 1.38, 95%CI 1.16-1.65) and in NSCLC patients (RR 1.51, 95%CI 1.12-2.03). There was no difference in de novo grade 3-4 iRAE and tumor response between cancer patients with and without AID. However, in NSCLC patients, pre-existing AID was associated with a 2-fold increased risk of de novo grade 3-4 iRAE (RR 1.95, 95%CI 1.01-3.75) but also better tumor response in achieving a complete or partial response (RR 1.56, 95%CI 1.19-2.04). Conclusions: Pre-existing AID confers an increased risk of toxicity during ICI therapy. NSCLC patients with AID are at a higher risk of de novo grade 3-4 iRAE but are also more likely to achieve treatment response than those without AID. Multidisciplinary collaboration is paramount when considering ICI therapy in this patient population with careful calculation of risk and benefit as well as close monitoring for toxicity.