BackgroundThe objectives of this review were to evaluate the vaccine effectiveness (VE) of the two-dose varicella vaccine for healthy children in China and explore the application of the approach of Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) in observational studies on VE.MethodsWe searched for observational studies on two-dose varicella VE for children in China aged 1–12 years that were published from 1997 to 2019, and assessed the quality of each study using the Newcastle Ottawa Scale (NOS). We used meta-analysis models to obtain the pooled two-dose VE, and the studies were divided into subgroups and analysed according to whether or not it was an outbreak investigation and its NOS score. The quality of evidence of VEs were rated by approach of the GRADE system.ResultsA total of 12 studies and 87,196 individuals were included. The pooled two-dose VE was 90% (95% confidence interval [CI]: 69–97%). The VE of outbreak studies (87% [95% CI: 76–93%]) was lower than non-outbreak studies (99% [95% CI: 98–99%]). There was no significant difference in VEs by different NOS quality. The quality of the evidence assessment of pooled two-dose VE was “low”, which was rated down by one category in limitations and publication bias respectively and rated up by two category in large effect. The quality of evidence assessment in subgroup of NOS score ≥ 7 was “moderate”.ConclusionsThe VE of two-dose varicella vaccine is relatively high in preventing varicella, and is recommended for countries which need further control for varicella. However, higher quality evidence is needed as a supplement for stronger recommendations. The approach of GRADE could be applied for rating the quality of evidence in observational study.