The article overviews the vicissitudes of the Saussurean notion of the two-sided linguistic sign in 20th-century linguistics. After having held centre stage for several decades in the Praguian and Copenhagian traditions of structuralism, this notion was sidelined if not rejected in the modular architecture of mainstream Generative Grammar, but has since been vindicated in the most influential linguistic schools to emerge in the wake of, and in reaction to, the Chomskyan revolution, viz. Cognitive Grammar and Construction Grammar. While the Saussurean theory of the linguistic sign may be claimed to be of foundational importance for the properly linguistic study of language, it was left for de Saussure’s successors to explore the implications and antinomies of the two-sided sign model. The article discusses a number of problems for the theory of the sign, such as the division between linguistically encoded and contextual meaning, the challenge of polysemy, the fluidity of the conventionalized linguistic sign in the continuous flow of diachronic processes, the Hjelmslevian dichotomy of semantic form and semantic substance, the place of syntax in a theory of the sign, etc.