Purpose: To compare the retention rates of non-carious cervical restorations (NCCLs) constructed using the sandwich technique (a lining of glass-ionomer cement [GIC] or resin-modified glass-ionomer cement [RMGIC] and composite resin [CR]) with CR-only restorations. Materials and Methods: The search was performed in various databases, including the Cochrane Library, PubMed, Scopus, and Web of Science. Gray literature was inspected, as were ongoing and unpublished abstracts from the IADR (1990-2017). Study quality was evaluated using the Cochrane Collaboration bias risk tool. Data from primary and secondary outcomes were meta-analyzed at 1-, 2- and 3-year follow-ups using the random effects model. The quality of the body of evidence was assessed using the GRADE approach. Results: Initially, a total of 3645 articles were selected. After selection by titles, abstracts, and full texts, 6 articles were retrieved, but three were follow-ups of the same RCT. Therefore, a total of four studies remained for analysis. All studies were at unclear risk for bias. Among all outcomes, only loss of retention was lower for the sandwich technique at the 3-year follow-up (risk ratio [RR]: 7.5; 95% CI: 2.1 to 27.2; p = 0.002). Conclusions: Based on the limited number of available studies, higher retention rates in NCCL restorations were observed with the sandwich technique compared to CR-only restorations at the 3-year follow-up. Secondary outcomes were not influenced by the restorative technique. Except for retention rates, which were of moderate quality, the evidence quality of all secondary outcomes was low. Keywords: glass-ionomer cements, permanent, dental restoration, dental materials, meta-analysis