On-chain forensic techniques enable probabilistic linking of Tornado Cash deposits and withdrawals despite unbroken zero-knowledge proofs.
Regulation & Gov ·
Tornado Cash employs zero-knowledge proofs to obscure the link between deposits and withdrawals, yet on-chain analysts have developed heuristic techniques to probabilistically connect transactions despite the cryptography remaining unbroken. These methods exploit behavioral patterns and metadata visible on the public blockchain rather than breaking the underlying cryptography. Researchers and compliance teams now use open-source tools to apply these forensic approaches at scale.
Demixing relies on multiple correlated signals extracted from transaction data. Analysts examine address reuse, where the same wallet both deposits and withdraws; gas price fingerprinting, matching unusual or identical gas prices between deposit and withdrawal transactions; timing correlation, identifying deposits and withdrawals occurring within narrow time windows; and fee patterns, since relayer withdrawals typically amount to slightly less than the fixed denomination. The tools tornado-demix and retrace automate these heuristics—the former scans for same-denomination vouchers deposited within configurable time gaps and searches for matching withdrawal counts, while the latter aggregates and indexes Tornado Cash and Railgun withdrawals for searchable analysis. These signals are probabilistic rather than definitive, generating leads requiring further verification.
Academic studies have shown these combined heuristics can link 20–35% or more of transactions together. What remains unclear is how widely these techniques are deployed operationally and whether new user behaviors might reduce their effectiveness.