Over the past decade, there has been increasing interest in marine restoration, requiring a consideration of various approaches for optimal success. Artificial reefs (ARs) have been employed for marine restoration and fisheries management, but their effectiveness in restoring ecosystems lacks well-defined ecological criteria and empirical evidence. A systematic review of the literature on ARs articles between 1990–2020, a meta-analysis of their effectiveness based on the similarity of species composition with reference natural reefs (NRs), as well as bias risk analyses were carried out. Research on ARs primarily focused production of marine communities (n = 168). There are important information gaps regarding socioeconomic aspects; design, materials, and disposal in the selected habitats; legal, management, and planning aspects considering long-term monitoring. Regarding effectiveness, few articles (n = 13) allowed comparisons between ARs and NRs, highlighting the need to apply proper reference sites in AR implementations. Meta-analysis showed that ARs are not similar to reference NRs (p = 0.03, common effect and p = 0.05 random effect models). However, a high index of heterogeneity (88%) suggests that this relation may be influenced by factors other than the reef type. Thus, further analysis can disguise variables conditioning this AR–NR similarity as a measure of restoration for degraded marine ecosystems.