The limit of detection is one of the fundamental metrological characteristics of any chemical measurement systems but, even now, it is a highly controversial matter because of the existing differences in terminology, formulations and understanding. Even when limiting the analysis to those operational approaches that, because of their simplicity, are often suggested by the guidelines of some international bodies, the available approaches can return appreciably different estimates. In this paper, the matter is critically reconsidered, and the different approaches compared with the aid of some case studies.