The notion of fitness, although fundamental in evolutionary theory, is not a unified concept. Several authors have even concluded that looking for a general concept of fitness is doomed to failure. I disagree with these conclusions. In the vast philosophical literature on fitness, two approaches have been put forward. One the one hand, under the statistical interpretation of the concept, fitness is the per capita rate of increase of a given type and does not play any causal role in evolutionary changes. On the other hand, under the ecological approach of the concept, fitness is a property of organisms (or more generally entities) forming a population in a given environment and it plays a causal role in evolutionary changes. While both approaches have some advantages over one another they also come with many drawbacks. In this paper, I propose a concept of fitness which takes the advantages of both approaches while eliminating their problems. I start by detailing different problem cases for each approach and propose “local” solutions for each of them. I then show that each solution can been integrated in a more general concept of fitness, yet consistent both with the idea that fitness is a property of entities and that it causally affects their evolutionary fate in the context of natural selection. I call this concept trait-based fitness. Further on, I show that trait-based fitness is perfectly equivalent to the notions of fitness used in classical evolutionary disciplines with the advantage of being more general.