The global losses of biodiversity and critical ecosystem services are great threats to human society. Ecological restoration is an important way to regain the lost ecosystem services and biodiversity. In order to achieve successful results, various restoration approaches have been developed. While some approaches take advantage of the self-designing ability of natural restoration systems, others rely on man-made designs at various levels. Since different restoration approaches may have different effects, identifying the best approach becomes crucial for successful restoration implementation. In order to achieve this objective, we proposed herein a system to classify various restoration methods into three types of approaches, namely, high-, intermediate-and low-intensity intervention approaches based on the intensity or the degree of human intervention. We then conducted a meta-analysis by using data obtained from ISI Web of Knowledge to study the effects of different restoration approaches on ecosystem services and biodiversity. In addition, we examined restoration effects under different climate zones, ecosystem types, restoration ages, and ecosystem service types. Finally, the relationship between biodiversity and ecosystem services was studied. We used the median response ratio as an indicator of biodiversity and ecosystem service restoration effects. Since our data were not normally distributed, Wilcoxon and Kruskal-Wallis non-parametric analyses were applied to detect statistical differences. Spearman rank analysis was used to test the correlation between biodiversity and ecosystem services. In this study, a low-intensity intervention approach indicates that the restoration effects are entirely achieved by natural forces. Human contributions to ecosystem restoration occur only through behavioral changes, such as land abandonment and threat removal. The intermediate-intensity approach applies when people only alter the environment during restoration. There is no direct intervention exerted on restoration targets. Some examples of intermediate intensity approaches include fertilizer application, the establishment of green corridors to improve habitat connectivity, and the addition of large deadwood to streams in order to improve habitat heterogeneity. A high-intensity approach is defined by direct human control on restoration targets. This approach usually involves anthropogenic biological recovery of a degraded ecosystem. Tree planting and species introduction are good examples of high-intensity approaches. The meta-analysis indicated that the median increment of ecosystem services and biodiversity in restored ecosystems was 45% and 151%, respectively, when compared to that of degraded ecosystems. The median enhancement of ecosystem services and biodiversity was 31% and 25%, respectively, for the low-intensity approach, and 31% and 22%, respectively, for the intermediate-intensity approach. A positive correlation was observed between biodiversity and ecosystem services, especially in restored versus degraded ecosystems. Further detailed analysis revealed a significant variation with regard to the effects of restoration approaches dependent on restoration goals (biodiversity or ecosystem services), referring systems (degraded or reference), climatic conditions, and time elapsed since restoration. However, some prominent differences were still found in tropical and terrestrial ecosystems. In this context, the high-intensity approach generally generated the best restoration effects when compared to degraded ecosystems. However, this approach may be suboptimal if the goal of restoration is to recreate the original environmental state. Indeed, the high complexity of reference ecosystems may be more easily recreated via intermediate intensity approaches. Our study emphasizes the importance of considering socioeconomic factors during restoration planning and creating a standard evaluation system for restoration effects and sustainability based on indicators of ecosystem services.