Background: There is an urgent need to improve health outcomes among severe COVID-19 patients. Therefore, this study aimed to demystify efficacy and safety of tocilizumab among those patients based on the best available evidence. Methods: Nine electronic databases were searched for relevant studies by three authors in July using search terms “tocilizumab”, “COVID”, “SARS”. The last date checked was 26 August, 2020. Eligible studies were assessed through an inclusion criterion focused on cohort studies and severe COVID-19 adult patients. Any disagreement between the authors was resolved by discussions. Primary outcomes were mortality and clinical improvement, while secondary outcomes were tocilizumab-related adverse effects. Results: Tocilizumab reduced the mortality rate by 41%, 40%, 28%, and 37% in 7, 14, 21, and 28 days. There was neither statistically significant difference between the tocilizumab group and the control group in the clinical improvement (pooled risk ratio1.21 [0.89, 1.64]) nor in respiratory support change. There were no difference between the two groups in the occurrence of serious adverse events (pooled risk difference 0.00 [-0.02, 0.03]), bacteremia (pooled risk ratio 1.25 [0.80, 1.97]), elevation of liver functions (pooled risk difference -0.00 [-0.03, 0.02]) and infusion related reactions (pooled risk difference 0.01 [-0.02, 0.03]). However, the occurrence of neutropenia (pooled risk ratio 9.23 [1.06, 80.24]) and new infections (pooled risk difference 0.07 [0.00, 0.14]) were more likely in the tocilizumab group. Conclusion: Because of the reported adverse effects of tocilizumab in treating severe COVID-19 patients, the use of that biological agent should be limited to selected patients based on the best available evidence.