Inhaled corticosteroid (ICS) in patients with chronic obstructive pulmonary disease (COPD) has been debated for 20 years. In our systematic literature review and meta-analysis, we addressed the following: Should patients with COPD and a blood eosinophil count (EOS) of, respectively, a) < 150 cells/μl, b) 150–300 cells/μl, and c) > 300 cells/μl continue treatment with ICS? Protocol registered in PROSPERO (CRD42020178110) and funded by the Danish Health Authority. We searched Medline, Embase, CINAHL and Cochrane Central on 22^nd July 2020 for randomized controlled trials (RCT) of ICS treatment in patients with COPD (≥40 years, no current asthma), which analyzed outcomes by EOS count and where >50% of patients used ICS prior. We used the GRADE method. Meta-analyzes for the outcomes were divided into EOS subgroups and analyzed for differences. We identified 11 RCTs with a total of 29,654 patients. A significant difference (p < 0.00001) between the three subgroups’ reduction of risk of moderate to severe exacerbation was found. Rate ratios for EOS counts: <150 cells/μL was 0.88 (95%CI: 0.83, 0.94); 150–300 cells/μL was 0.80 (95%CI: 0.69, 0.94); >300 cells/μL was 0.57 (95%CI: 0.49, 0.66). Overall, the certainty of the effect estimates was low to very low due to risk of bias, unexplained heterogeneity, few RCTs, and wide confidence intervals. A clear correlation was demonstrated between effect of continued ICS treatment (number of exacerbations, lung function, and quality of life) and increasing EOS count. Our meta-analyses suggested that treatment with ICS seemed beneficial for everyone except patients with EOS count below 150 cells/μl.