Understanding the ecological mechanisms that constrain forest succession in tropical degraded anthropogenic grasslands is a prerequisite for the design of techniques for restoring biodiversity and ecosystem processes. In this context, succession on post-agricultural lands may be arrested by a variety of site-specific biotic and abiotic factors. Here we synthesise our research on the effects of five biotic factors (seed dispersal, development of a soil seed bank, seedling emergence, herbivory, competition) and five abiotic factors (fire, microclimatic conditions, soil nutrients, water availability, disturbance) as constraints to forest succession on degraded anthropogenic grasslands in a tropical lower montane forest landscape in Sri Lanka. The aim of this research was to deduce ecologically and socially acceptable restoration techniques to accelerate forest recovery. Colonisation of grasslands by trees is constrained by limited seed dispersal from adjacent remnant forest patches and their incorporation into grassland soil seed banks. For the few tree seeds that are dispersed into grasslands, a combination of vertebrate herbivory and annual dry season fires reduces the likelihood that they emerge as seedlings. Removal of the grass canopy by clipping or tilling increases the emergence of woody plant seedlings close to the boundaries of forest patches, but has no effect beyond 20 m into the established grassland. Our research shows that isolation of seedling root systems from those of competing grasses increases the growth and survival of tree seedlings transplanted directly into grassland swards, while above-ground competition and exclusion of vertebrate herbivores has no effects on seedling growth and survival. These experiments identified that the early-successional species Macaranga indica Wight and Symplocos cochinchinensis (Lour.) S. Moore are potential candidates for use in reforestation programmes on abandoned grasslands. We propose a strategy for a model forest restoration programme based on the creation of vegetation islands using early-successional native tree species, the application of a tilling treatment around remnant forest patches, creation of fire breaks around vegetation islands, and the protection of isolated individual trees and tree patches within established grasslands. We highlight the importance of further research on the ecology and biology of seed dispersers and seed predators, and expansion of knowledge on the regeneration traits of native tree species, for future refinements of this restoration strategy. [ABSTRACT FROM AUTHOR]