Since the 1986, the National Institute for Occupational Accident Insurance (INAIL) provides each single Region with the occupational accident data on magnetic support. This strengthens the need to critically evaluate the most commonly used risk indicators. In this paper we discuss advantages and limits of the usual frequency and severity measures; the formulae for the standard errors are also given. A method proposed by I. Bross (1958), the RIDIT Analysis, is considered and exemplified. The appealing properties of this method ask for a more extensive application in this field.