Comparisons between putatively neutral genetic differentiation amongst populations, F(ST), and quantitative genetic variation, Q(ST), are increasingly being used to test for natural selection. However, we find that approximately half of the comparisons that use only data from wild populations confound phenotypic and genetic variation. We urge the use of a clear distinction between narrow-sense Q(ST), which can be meaningfully compared with F(ST), and phenotypic divergence measured between populations, P(ST), which is inadequate for comparisons in the wild. We also point out that an unbiased estimate of Q(ST) can be found using the so-called 'animal model' of quantitative genetics.