In several places we have argued that ‘fitness’ is a primitive term with respect to the theory of evolution properly understood (Rosenberg 1978, 1982, 1983; Williams 1976). These arguments have relied heavily on the axiomatization of the theory provided by one of us (Williams 1970). In contrast, both John Beatty and Robert Brandon have separately argued for a “propensity“ interpretation of “fitness” (Mills and Beatty 1979; Brandon 1978); and in Brandon and Beatty (1984) they attack our view that “fitness“ is a primitive term in evolutionary theory, concluding that a definition by way of propensities is possible and preferable. Here we reply to their criticisms, and argue that, at most, the view that fitness is a statistical propensity is a terminological variant on our thesis that it is a primitive in evolutionary theory.