The dominant protocol to study the effects of plant diversity on ecosystem functioning has involved synthetically assembled communities, in which the experimental design determines species composition. By contrast, the composition of naturally assembled communities is determined by environmental filters, species recruitment and dispersal, and other assembly processes. Consequently, natural communities and ecosystems can differ from synthetic systems in their reaction to changes in diversity. Removal experiments, in which the diversity of naturally assembled communities is manipulated by removing various components, complement synthetic-assemblage experiments in exploring the relationship between diversity and ecosystem functioning. Results of recent removal experiments suggest that they are more useful for understanding the ecosystem effects of local, nonrandom extinctions, changes in the natural abundance of species, and complex interspecific interactions. This makes removal experiments a promising avenue for progress in ecological theory and an important source of information for those involved in making land-use and conservation decisions.