Decomposition of plant detritus by soil microbes is a key ecosystem process which determines the magnitude of ecosystem carbon storage. Although tropical forests play a crucial role in regulating the global carbon cycle, controls over decomposition in these forests are not well understood. The combination of constant warm temperatures, extremely high precipitation inputs, and old, highly leached soils characteristic of such ecosystems is not accurately captured by current models developed for temperate forests. Furthermore, spectacular plant chemical diversity complicates efforts to identify key drivers of leaf litter decomposition in tropical forests. For instance, recent studies in three separate sites have found that the best predictor of decomposition rate is foliar tannin concentration (Coq et al 2010) vs. leaf lignin:P ratios (Wieder et al 2009) vs. availability of micronutrients (Kaspari et al 2008), respectively. Therefore, this session aims to synthesize hypotheses developed by multiple researchers in order to achieve a more mechanistic framework for understanding leaf litter decomposition in the tropics. Following the hierarchical model of abiotic and biotic controls on decomposition developed by Lavelle et al. (1993), the session will begin with a discussion of the large-scale drivers of decomposition rates: temperature, precipitation regime, and soil mineralogy. Speakers will then address chemical controls on decomposition, with particular attention to the roles of foliar nutrient content vs. non-lignin foliar carbon. Finally, presenters will discuss the sensitivity of decomposition to composition of mesofaunal, bacterial, and fungal communities in soil. The experiments discussed will range in scale from controlled laboratory microcosm studies to pan-tropical analyses of patterns observed across the entire tropical forest biome. At the conclusion of the session, speakers will discuss how we might reconcile the many mechanistic hypotheses of leaf litter decay in tropical forests, and how such a new predictive framework might be incorporated into global carbon cycling models.