This dissertation is a study of the De Signis of the Epicurean philosopher Philodemus of Gadara. In this treatise, he provides four accounts of how certain Epicureans defended their views on signs and sign-inference against the attacks of a group of opponents. I reconstruct and critically discuss these views on the basis of what is said about them in attack and defence. Two so-called 'methods of sign-inference' provide tests that allow one to establish that certain inferences are valid and certain conditionals true. The method of similarity applies to certain kinds of generalising inference, and is regarded as primary; the method of elimination, which licenses inferences from the existence of some phenomenon to that of an underlying state that is required by that phenomenon and explanatorily prior to it, is taken to be somehow dependent on the method of similarity.