The past two decades have witnessed an expansion in efforts to publicly disseminate data on hospital performance based on comparisons of risk-adjusted outcomes for the purpose of affecting reimbursement or patient choice. While much is known about which risk factors should be adjusted for, less is known about the appropriate statistical methods that should be used in deriving such quality measures. We discuss the literature on profiling and risk adjustment, with an emphasis on recent econometric and statistical methods, highlighting key assumptions involved in the various analytical techniques. Particularly problematic for the traditional methods of analysis are inadequate sample sizes and unobserved severity of illness. We illustrate how these issues affected recent public profiling initiatives and highlight how recent contributions from the econometric and statistical literature may be helpful in ameliorating these problems.