Striking progress has been made on conceptual and methodological aspects linking species traits to community and ecosystem responses to environmental change. However, the first step when using a trait-based approach (i.e., choosing the adequate traits reflecting species response to a given environmental driver) deserves much more attention. The first broad comparative studies, using worldwide datasets, have identified a number of traits that are, for example, good indicators of plant responses to water and nutrient availability, or good indicators of plant species defense against herbivory and plant effects on litter decomposition. Due to the successful explanation of global patterns of trait variation and the relatively easy measurements, some functional traits have become widely used. However, it is starting to be questioned whether the status of such "fashionable traits" is always justified; especially considering that particular traits might be the result of underlying traits that respond to distinct environmental factors. Some global trade-offs between traits reflecting contrasting resource use strategies do not hold under changing environmental conditions. Therefore, the choice of traits to up-scale responses of individuals to communities and ecosystems should be based on their adequacy for a specified ecological context. Here, we present a framework that helps identifying objective criteria for the choice of functional traits for studies on community assembly and ecosystem processes and services. This framework comprises five steps: 1) identification of the main environmental drivers; 2) identification of the relevant physiological processes allowing species to cope with the environment; 3) selection of traits that are involved in such physiological processes; 4) validation of the select traits at community level and; 5) evaluation of possible consequences for ecosystem processes/services.