This paper critically reviews the ontological debates over the nature of ethnicity and the different ways in which it is operationalized and “measured” for quantitative research. It is argued that while moving away from a “primordealist” position on ethnicity renders measurement of the social diversity more difficult, conceptually and practically it does not invalidate this exercise. A second problem, however, is also identified with the measurement of ethnicity: when information on ethnic diversity is incorporated with other socio-economic information, a range of measures can be derived that purport to pick up very different distributions, but that are in reality often very highly correlated. These two problems combined present a significant challenge for the quantitative study of the relationship between ethnic diversity and political and economic outcomes such as conflict and growth patterns. The authors do not assert that these problems invalidate the exercise of investigating these relationships econometrically entirely, but they suggest the problems do warn us to be more guarded and modest in the claims made on the basis of such analyses.