Objectives: This paper examines the predictors of individual variations in punitivity: what makes some individuals or countries more punitive than others? Methods: We differentiate between expressive and considered punitivity. We analyse to what extent perceptions of justice system legitimacy, measures of right-wing orientation (using Derex scales), and instrumental motivators can help explain attitudes to punishment, using data from Round 5 of the ESS. Although initially we envisaged a multi-level modelling approach, assessing the relative weight of individual and country-level predictors, an insufficient proportion of the variance in public punitivity could be attributed to country-level to permit this. We brought down the analysis to a single level and pooled countries to identify individual-level predictors using a binary logit model and an ordered logit model with partial proportional odds assumption. Results: Based on our models right-wing value orientation and views about legitimacy appear to be the strongest predictors of both types of punitivity while instrumental motivators seem to have a secondary role only. Conclusions: The lack of cross-national variation in attitudes to punishment highlights the need to consider what can be deemed as culturally rooted in criminology. The results provide support for a Neo-Durkheimian value-expressive model of subjective punitivity.