The principal purpose of automatic signal classification is resource allocation. An incoming signal is quickly classified by “front-end” processing to determine which “follow-on” processor(s) to apply to it. The allocated processor could be a computer, a modem, a demultiplexer, diagnostic equipment, and so on. This paper introduces each of the major steps in developing a signal classifier: ascertaining distinctive signal characteristics; deriving signal “features,” statistics that capture signal characteristics; determining the most powerful features within computational constraints; selecting a decision logic; and logical integration of classification decisions.