Abstract Notwithstanding the general acceptance of the value of statistical life (VSL) estimates for policy assessment purposes, several important unresolved issues remain. First, the results from revealed preference studies are systematically higher than those from stated preference studies, potentially limiting the usefulness of stated preference studies in generalizing the VSL estimates to different populations and kinds of risks. Second, extrapolating the results of meta-analyses to project the VSL for different population groups requires that such generalization be reflective of the underlying economic content of what average VSL estimates reflect. Third, government agencies within and across countries place differing emphasis on types of VSL studies as well as differing reliance on individual studies versus meta-analyses. Usually, there is no justification provided for the chosen approach.