Starting from Adam Smith's intuition, compensating wage differentials are one of the most widespread explanation to describe why agents should bear occupational risk of injury and death. For nearly thirty years, economists have attempted to and empirical evidence on such wage differentials mostly relying on estimation of a simple wage equation. This paper claims to put one step forward. Using the Survey of Household Income and Wealth (SHIW) 2004 we estimate for Italy the wage premium held by workers in risky occupations by means of the matching estimator. Such technique is desirable because it attempts to remove all the differences in wage coming from heterogeneity across individuals and not directly imputable to risk. Estimates suggest that net hourly wage premium is about 3% to manual workers and nearly null to non-manual workers. When we split the sample along the employer size, our findings show a heterogeneous treatment with respect to occupational status. Small firms tend to flatten out any risk premium to manual workers, while they recognize roughly 6% to non-manual workers; the opposite occurs when we look at medium-large firms wherein manual workers gain 1.5% to 5% more with respect their counterparts. Therefore, it seems that wage-risk trade off does not always emerge as hedonic wage theory would predict.