The aim of this paper is to explore possible measures to assess welfare changes resulting from alternative risk communication policies drawing on previous work on welfare losses due to incorrect risk perceptions. A review of the literature analysing models of risk information processing point out the role of Bayesian updating processes in modelling changing risk perceptions. On the other hand welfare analysis of the consequences of action undertaken under ignorance suggest possible measures of the benefits of improved risk communication. An illustrative example of welfare losses due to poor information is provided using the classical data set on milk contamination in Ohau (Hawaii). Welfare measures are obtained comparing the real observed behaviour with an hypothetical risk perception pattern induced by a more effective risk communication policy.