<p>The self-interest approach to preferences for redistribution draws upon the idea that socio-economic conditions influence policy preferences via personal interests. Following this principle, a burgeoning interdisciplinary literature has examined the influence of socio-economic status (SES) on redistributive preferences. Yet this body of research is quite fragmented because it includes a plethora of understandings of SES and there is still no consensus on which conceptualization best accounts for variation in these preferences. We fill this gap in the literature through an analysis of the predictive validity of seven conceptualizations of SES: (i) income as a linear measure; (ii) income measured in deciles; (iii) skills specificity; (iv) ESeC schema; (v) Oesch class schema; (vi) risk of unemployment; and (vii) routine task intensity. Using data from the European Social Survey for 24 countries in 2012–2022, we determine the predictive power of each conceptualization through different measures of goodness of fit. The results show that income measured in deciles is the conceptualization with highest predictive power and parsimony.</p>