Over many decades, modelling natural communities of higher diversity and complexity has been hampered by the necessity to provide reasonable parameters for all processes at the level of populations and interactions. Trait-based approaches such as allometric scaling allow to use general species traits such as their average individual body mass to estimate the parameters of populations (e.g., metabolic rates) and interactions (e.g., attack rates). This chapter describes this trait-based network approach and illustrates its potential for understanding the structure and dynamics of complex networks using the examples of i) intrinsic community stability and ii) consequences of global change (e.g., warming). Finally, new research frontiers are illustrated that include spatial processes in meta-networks, the constraints of ecological network structure on ecosystem functioning, and non-trophic interactions.