<p>This paper develops and evaluates a deterministic entity-resolution framework for matching shadow-fleet tanker records across heterogeneous maritime datasets under conditions of identifier suppression, manipulation and cross-source inconsistency. While the literature on Russian sanctions evasion, AIS manipulation and shadow-fleet risk has grown rapidly, it still lacks a transparent and reproducible method for establishing vessel correspondences when canonical identifiers are unavailable or unreliable. The model proposed here addresses that gap through a weighted, multi-pass, one-to-one matching architecture based on year of build, deadweight tonnage, flag and vessel class. The results show a deliberately conservative but highly precise matching regime, with strong performance on validation and a clear precision–recall trade-off. The analysis demonstrates that ambiguity control, especially through best-versus-runner-up margins, is more important than simple permissive matching. The paper contributes a replicable identification layer for subsequent sanctions, compliance, intelligence, legal and maritime-risk analysis of the shadow fleet.</p>