We provide more evidence on the functional relationship between willingness-to-pay for risk reductions and age (the senior discount). We overcome many of the limitations of previous literature that has dealt with this issue, namely, the influence of the assumptions used in statistical models on the final results. Given our large sample size (n = 6024) we can use models that are very demanding on data. We use parametric (linear, quadratic, dummies), semi-nonparametric, and non-parametric models. We also compare the marginal and the total approach and show that they provide similar results. We also overcome one of the limitations of the total approach, that is, we include the effects of socioeconomic characteristics that are correlated with age (education and income). Our main result is that all these different approaches produce very similar results, namely, they show an inverted-U relation between the value of a statistical life (VSL) and age. Those results can hardly be attributed to problems of collinearity, omitted variables or statistical assumptions. We find a clear senior discount effect. This effect seems concentrated on those who have lower education and income levels. We also find that the value of a statistical life year (VSLY) increases with age.