Aim:An increasing number of studies have reported a higher frequency of liver injury in COVID-19. Several meta-analyses of the associations between liver injury and COVID-19 had been published, but the dose-response relationship has not been conducted. Therefore, we performed a dose-response meta-analysis to explore this relationship. Methods: PubMed, Web of Science, EMBASE, and four additional Chinese databases were used to collect studies from December 1, 2019 to October 26, 2020. Key information was extracted, and the corresponding model was selected on the basis of the heterogeneity test. The pooled RR, OR, and WMD values were then calculated. Sensitivity analysis and publication bias were also assessed to further demonstrate the stability and reliability of the results. Results: Among the 13298 studies that were initially retrieved, 54 studies met our inclusion criteria and included 9907 participants. There was a dose-response relationship between ALT (OR from 1.0 for ≤40 U/L to 1.556[1.130–2.142] for 40–80 U/L and 1.635[1.176–2.273] for >80 U/L), AST (OR from 1.0 for ≤40 U/L to 1.635[1.180–2.266] for 40–80 U/L and 2.019[1.387–2.938] for >80 U/L) and severe COVID-19. Higher ALT, AST, and TBil were associated with need for ICU admission (WMD 6.45[4.08–8.81], 8.80[6.02–11.57], 2.03[0.91–3.14]), and death due to COVID-19 (WMD 3.22[0.08–6.35], 12.06[6.26–17.85], 2.70[0.72–4.68]). There was no publication bias in this meta-analysis. Conclusion: There were dose-response relationships between ALT, AST, and COVID-19 severity. ALT and AST levels in COVID-19 patients on admission should be considered to be clinical warning indicators, which could provide basis for preventing serious COVID-19 and associated deaths