Recent philosophical discussions have failed to clarify the roles of the concept fitness in evolutionary theory. Neither the propensity interpretation of fitness nor the construal of fitness as a primitive theoretical term succeed in explicating the empirical content and explanatory power of the theory of natural selection. By appealing to the structure of simple mathematical models of natural selection, we separate out different contrasts which have tended to confuse discussions of fitness: the distinction between what fitness is defined as versus what fitness is a function of, the contrast between adaptedness as an overall property of organisms and specific adaptive capacities, the distinction between actual and potential reproductive success, the role of chance versus systematic causal relations, fitness as applied to organisms as opposed to fitness applied to genotype classes, heritable adaptive capacities of genotypes as opposed to relations between genotypes and the environment. We show how failure to distinguish and properly interrelate these different aspects of “fitness” adds confusion to a number of already complex issues concerning evolutionary theory. On the basis of our discussion of these different aspects of “fitness”, we propose a terminology which makes the necessary distinctions. A central result of our analysis is that the concept of fitness as the overall adaptedness of organisms does not enter into the causal structure of evolutionary explanation, at least to the extent that this structure is represented in the mathematical models of natural selection.