This dissertation develops a comprehensive theory of correlative constructions in ancient Indo-European languages, challenging the traditional view that correlatives are a subtype of relative clauses. Focusing on Vedic Sanskrit, Old and Young Avestan, Old Persian, Hittite, and Homeric Greek, the study presents both synchronic and diachronic analyses to illuminate the syntactic behavior and historical evolution of correlativity. Synchronic analysis reveals that correlative constructions in these languages exhibit significant structural differences from modern correlatives (e.g., in Hindi), particularly in terms of headedness, clause position, and syntactic integration. Using formal syntactic frameworks, the dissertation argues that Vedic correlatives are base-generated in the left periphery of the host clause and display a range of headedness types, including internally headed, externally headed, double-headed, free, and split-headed variants. These findings contest the prevailing typology that assumes correlatives must be internally headed. The diachronic portion of the dissertation reconstructs correlative constructions in Proto-Indo-European by tracing the grammaticalization paths across early attested Indo-European branches. It is proposed that the evolution from raised headed (cor)relatives to embedded relative clauses involved morphological reanalysis to the precursor of the izafe construction (“prezafe”) Stage II, i.e. nominal relative clauses with case attraction. Ultimately, the dissertation redefines the role of correlatives in syntactic theory by demonstrating their independence from standard relativization strategies. It introduces a multidimensional classification framework based on the relative-correlative distinction, headedness, and linear order. This work contributes to our understanding of clause structure, the typology of relativization, and the historical syntax of Indo-European languages