The value of statistical life (VSL) is one of the most scrutinized and controversial parameters estimated by environmental economists (Cameron 2009, Viscusi 2010), largely due to the wide use of VSL estimates to value the mortality risk benefits of regulations that affect public health and safety (OMB 2011, Robinson and Hammitt 2010). The hedonic wage method has been a primary source of VSL estimates for use in applied benefit‐cost analysis and there have been several meta‐analyses of these studies, including examinations of publication bias. We build on the existing literature by focusing on the coefficient on fatal risk rather than the VSL itself. This allows for larger sample sizes and reflects more recent methods that provide a cleaner test for bias. Results suggest that publication bias is present in the full sample of hedonic wage VSL estimates and that correcting for this by using those observations with the most precise estimates results in lower mean VSL estimates.