Background: Neutrophil-to-lymphocyte ratio (NLR) is an accessible and widely used biomarker. NLR may be used as an early marker of poor prognosis in patients with COVID-19. Methods: We conducted a systematic review and meta-analysis. Observational studies that reported the association between baseline NLR values (i.e. at hospital admission) and severity or all-cause mortality in COVID-19 patients were included. The quality of the studies was assessed using the Newcastle-Ottawa scale (NOS). Random effects models and inverse variance method were used for meta-analyses. The effects were expressed as odds ratios (OR) and their 95% confidence intervals (CI). Small study effects were assessed with the Egger’s test. Results: Twenty studies, 19 cohorts and one case-control were included. An increase of one unit of NLR was associated with a higher odds of COVID-19 severity (OR 6.6, 95% CI: 4.71 - 7.19; p<0.001) and higher odds of all-cause mortality (OR 12.7, 95% CI: 1.32, 123.36; p=0.025). No differences were found in subgroup analyses by study design. The subgroup analysis of the studies, by country of origin, showed that the strength of the association between NLR and mortality was greater in Chinese studies (OR 31.1; 95%CI 19.57 to 49.3; p<0.0001) with moderate heterogeneity (I2 =43%). In our sensitivity analysis, we found that 7 studies with low risk of bias maintained strong association between both outcomes and the NLR values (severity: OR 4.7; 95% CI 3.5 to 6.34; p < 0.001 vs mortality: OR 31.1; 95% CI 19.57 to 49.3; p <0.0001), with low (I2 = 37%) and moderate (I2 = 43%) heterogeneity for severity and mortality outcomes, respectively. No publication bias was found for studies that evaluated effects for the severity of disease. Conclusions: Higher values of NLR were associated with severity and all-cause mortality in hospitalized COVID-19 patients.