Background Enhancing water provision services is a common target in forest restoration projects worldwide due to growing concerns over freshwater scarcity. However, whether or not forest cover expansion or restoration can improve water provision services is still unclear and highly disputed. Purpose The goal of this review is to provide a balanced and impartial assessment of the impacts of forest restoration and forest cover expansion on water yields as informed by the scientific literature. Potential sources of bias on the results of papers published are also examined. Data sources English, Spanish and Portuguese peer-review articles in Agricola, CAB Abstracts, ISI Web of Science, JSTOR, Google Scholar, and SciELO. Databases were searched through 2015. Search terms Intervention terms included forest restoration, regeneration/regrowth, forest second-growth, forestation/afforestation, and forestry. Target terms included water yield/quantity, stream-flow, discharge, channel runoff, and annual flow. Study selection and eligibility criteria Articles were pre-selected based on key words in the title, abstract or text. Eligible articles addressed relevant interventions and targets and included quantitative information. Results Most studies reported decreases in water yields following the intervention, while other hydrological benefits have been observed. However, relatively few studies focused specifically on forest restoration, especially with native species, and/or on projects done at large spatial or temporal scales. Information is especially limited for the humid tropics and subtropics. Conclusions and implications of key findings While most studies reported a decrease in water yields, meta-analyses from a sub-set of studies suggest the potential influence of temporal and/or spatial scales on the outcomes of forest cover expansion or restoration projects. Given the many other benefits of forest restoration, improving our understanding of when and why forest restoration can lead to recovery of water yields is crucial to help improve positive outcomes and prevent unintended consequences. Our study identifies the critical types of studies and associated measurements needed.