Forest restoration has the potential to mitigate the impact of deforestation and forest degradation.Various global policies have been sought to put restoration into the mainstream agenda including under the Convention on Biological Diversity (CBD) and the program for Reducing Emissions from Deforestation and forest Degradation (REDD+). The Aichi Target of the CBD set a target for at least 15% of degraded ecosystems to be restored by 2020 for key goals including biodiversity conservation, carbon enhancement and the provision of livelihoods. A theoretical framework to underpin decisionmaking for landscape-scale restoration has been slow to emerge, resulting in a limited contribution from science towards achieving such policy targets. My thesis develops decision frameworks to guide the restoration of degraded tropical forests to enhance biodiversity and the delivery of ecosystem services. In this thesis, three critical questions on how to make better decisions for landscape-scale restoration are addressed by: (a) considering landscape heterogeneity in terms of degradation condition, restoration action and cost, and temporally-explicit restoration benefits; (b) leveraging restoration within competing land uses using emerging policy for offsetting; and (c) enhancing feasibility by accounting for the social and political dimensions related to restoration.I use Kalimantan (Indonesian Borneo) as a case study area, as it represents a region that is globally important in terms of biodiversity and carbon storage. Kalimantan's forests also provide essential livelihoods for local people. Yet, rapid deforestation and forest degradation threaten the forests in this region. Chapter 2 verifies that forest loss and degradation is the most significant threat to biodiversity in Kalimantan, impacting more than 80% of threatened animal species and 60% of threatened plant species. The future of Kalimantan's wildlife depends on the survival of species in human-modified landscapes including in restored forest.Quantifying carbon benefit in a policy, such as forest restoration under the REDD+, requires a standardised tool, which has not been available for data-poor regions including Kalimantan. In Chapter 3, I examine a process-based model, called 3-PG (physiological principles for predicting growth), to estimate the above-ground biomass (AGB) content of the major forest types occurring on the island of Borneo. Using readily available climate and soil data, the results indicate the 3-PG model accurately predicts AGB compared with field-measured data, revealing the potential application of this model for carbon sequestration analyses. The datasets along with a set of parameters used in this chapter are employed in the subsequent chapters.Degraded tropical forests are characterised by a broad spectrum of forest condition states, mainly as a result of varying intensities of logging and fire. In Chapter 4, I develop a new framework to ii optimally allocate restoration investments to forests of varying condition, with two contrasting o...