Subjective well-being (SWB) data is increasingly used to perform welfare analyses. Interpreted as 'experienced utility', SWB has recently been compared to 'decision utility' using specific experiments, most often based on stated preferences. Results point to an overall congruence between these two types of welfare measures. We question whether these findings hold in the more general framework of non-experimental and large-scale data, i.e. the setting commonly used for policy analysis. For individuals in the British household panel, we compare the ordinal preferences either "revealed" from their labor supply decisions or elicited from their reported SWB. The results show striking similarities on average, reflecting the fact that a majority of individuals made decisions that are consistent with SWB maximization. Differences between the two welfare measures arise for particular subgroups, lending themselves to intuitive explanations that we illustrate for specific factors (health and labor market constraints, 'focusing illusion', aspirations).