Research
Reuse of Public Keys Across UTXO and Account-Based Cryptocurrencies
Address reuse is known to harm privacy, and reuse across different cryptocurrencies has been observed before, but prior work could only detect it where address formats were identical or directly…

- arXiv
- 2601.19500
- Published
- 2026-01-27
- Authors
- Rainer Stütz, Nicholas Stifter, Melitta Dragaschnig, Bernhard Haslhofer, Aljosha Judmayer
AI summary
Overview
Research area: Cryptocurrency privacy, blockchain forensics, and cross-chain entity linking (cs.CR). Technical level: Intermediate — assumes familiarity with UTXO versus account-based blockchain designs and ECDSA/address derivation, but the core argument is accessible. Scope: A measurement study of secp256k1 public key reuse within and across six cryptocurrencies (Bitcoin, Ethereum, Litecoin, Dogecoin, Zcash, Tron), using extracted public keys rather than addresses to link entities across incompatible address formats.
What This Paper Is About
Address reuse is known to harm privacy, and reuse across different cryptocurrencies has been observed before, but prior work could only detect it where address formats were identical or directly convertible. Because Bitcoin-style (UTXO) and Ethereum-style (account-based) systems derive visually incompatible addresses from the same secp256k1 keys, cross-design reuse has remained unexplored. The authors instead compare the underlying public keys themselves, which lets them match keys across Bitcoin, Ethereum, Litecoin, Dogecoin, Zcash and Tron, and quantify how much key material is shared between fundamentally different system designs.
Key Contributions
- A large-scale measurement showing extensive and ongoing key reuse across all six analyzed networks: 1,604,614 reused keys, of which at least 1,429,008 were actively reused (used for signing in more than one cryptocurrency).
- The first cross-chain key reuse analysis bridging UTXO and account models, reporting 497,178 keys shared between Bitcoin and Ethereum and 831,056 in total between all analyzed UTXO and account-based systems — roughly 50% of all observed reused keys.
- A non-heuristic clustering method that transfers UTXO-style clustering (such as the multiple-input heuristic) into account-based systems by deriving candidate addresses in the account-based chain from public keys observed in the UTXO chain.
- Validation of the key-based approach against prior address-based work (HSI and KeyLinker/[SY25]) and against an Ethereum–Tron ground truth built from their largely compatible address formats.
Main Findings
- Scale of reuse: Across BTC, ETH, LTC, DOGE, ZEC and TRX, 1,604,614 keys appear in at least two systems, and at least 1,429,008 were actively reused. Keys are reused "extensively and continually."
- Cross-design reuse is substantial: 497,178 keys are shared between Bitcoin and Ethereum, and 831,056 keys link UTXO and account-based systems — about half of all reused keys observed.
- A phenomenon of the present, not just the past: Quarterly counts for Bitcoin and Litecoin show internal reuse continuing through the analyzed period (data ends with blocks mined just prior to April 1, 2025).
- Internal UTXO reuse patterns: The P2PKH-P2WPKH pair accounts for 44.6% of internal reuse events and P2WPKH-P2PKH for 32.9%, together roughly 77.5%; the remaining pairs account for 22.5%. Reuse involving compressed versus uncompressed keys (P2PKH-P2PKH) is only 1.8% of internal events, and 1.99% for Bitcoin alone.
- Very low internal reuse in some chains: Zcash shows a total of four instances of internal public key reuse; Dogecoin has a median of two first active internal reuse events per quarter, peaking at 168 events in the first quarter of 2018.
- Cross-chain concentration: The six most frequent currency pairs, all involving BTC, ETH and TRX, constitute approximately 73.3% of all first active cross-chain reuse events. The TRX-ETH pair alone accounts for 23.3%, and keys more often appear first in Tron than in Ethereum. The remaining pairs ("Other") make up 26.7%.
- Clustering gains: Applying key-based merging on top of multi-input heuristic clusters merged 2,040,417 clusters, reducing the total from 618,812,655 to 616,772,238.
- Cross-design clustering output: The method grouped 497,289 Ethereum addresses into 62,681 clusters and 305,019 Tron addresses into 19,242 clusters.
- Services are involved, not only individuals: 1,200 addresses from 894 unique key-reuse instances map to two exchanges ranked among the Top 10 spot exchanges by volume on CoinMarketCap, accounting for 792 and 102 reused public keys respectively. About 26% of these exchange-linked instances involve higher-order intersections spanning three or more blockchains.
- Other attributed categories: 1,297 reused keys associate with Name Services, 1,070 with Contract Deployers, 911 with DeFi Bridges, and 203 with Mixing services.
- Weak keys are not the explanation: 279 weak private keys (a low number below 2^32, or a 256-bit string of Hamming weight below 5) were found among reused keys, but 88% of the ETH keys still hold value, collectively controlling more than 1 million Ether.
- Validation results: Against HSI's cases within their block range, the key-based method identified 59% for Bitcoin-Litecoin, 89% for Dogecoin-Bitcoin and 93% for Dogecoin-Litecoin — the shortfall being exclusively passive reuse that public-key extraction cannot see. Extending to the current block range, the ratios were 89% for Bitcoin-Litecoin, 200% for Bitcoin-Dogecoin and 300% for Litecoin-Dogecoin. Against the Ethereum–Tron ground truth, 91% of reuse events were identified.
- Likely cause: Deterministic (HD) wallet derivation without strict domain separation. Wallets such as MetaMask and Exodus derive keys from a single coin_type branch and let users "switch networks" while keeping the same keys; Tron, by contrast, defines its own SLIP44 coin_type and uses a different derivation path.
Methodology in Plain English
Public keys are normally hidden behind hashed addresses, but they are revealed when funds are spent, because a spending transaction must carry a signature that exposes the key for verification. The authors therefore reconstructed the actual secp256k1 public keys behind transactions rather than working with address strings.
For UTXO chains they used a modified version of bitcoin-etl (version 1.5.2) with added SegWit fields, and handled each output script type separately to recover keys. For Ethereum they parsed all transactions initiated by externally owned accounts (EOAs) and recovered keys with a signature-recovery method, since the Geth node does not expose keys directly; this produced 226,017,011 unique public keys. For Tron, given the volume of transactions, they first isolated 266,333,226 transactions where a new source/owner address first transacted, then recovered keys for those.
They then treated each cryptocurrency's key set as a set and computed intersections, visualized with an UpSet plot. Crucially, they separated active reuse (the key signed a transaction in at least two chains — requiring the private key, hence strong evidence of one controlling entity) from passive reuse (the key was only a recipient elsewhere, which the owner cannot prevent). Finally, because one public key maps to many possible addresses, they used a key seen in a UTXO chain to derive candidate addresses in an account-based chain, merging clusters across designs.
Why This Matters
The paper shifts cross-chain analysis from address-string matching to the cryptographic keys themselves, which yields entity links that are not products of heuristic guesswork. This gives researchers a ground-truth baseline for evaluating existing clustering heuristics, and a way to extend UTXO-only techniques into account-based systems where the multiple-input heuristic has no direct analogue.
Real-world applications:
- Blockchain forensics and tracing: Linking addresses of the same controller across chains, including through mixing services, where 203 reused keys were attributed to mixing services — creating cross-chain identifiers even in flows designed to obfuscate.
- Exchange and service compliance: Identifying that top-10 spot exchanges reuse keys across many networks helps explain patterns in cross-chain transaction monitoring and attribution.
- Wallet design and security guidance: Since a single compromised private key would drain funds across every chain where it was reused, the findings argue for per-cryptocurrency key derivation and domain separation in HD wallets.
- Privacy tooling and user awareness: Documenting that popular EVM wallet software encourages cross-network key reuse, which reduces unlinkability and enables deanonymization.
Industry relevance is direct: the measurements implicate major exchanges, DeFi bridges, name services and contract deployers, and the authors note that bridge designs such as the trusted-execution-based Avalanche Bridge intentionally reuse keys between UTXO and account models for usability, with observable spill-over into Ethereum.
Future Directions
- Determining conclusively whether cross-chain reuse between UTXO and account-based designs is intentional or inadvertent, and verifying whether HD wallet reuse between these designs is present in the dataset — the authors call this difficult to verify reliably and leave a detailed investigation to future work.
- Extending public key extraction to smart contract input data, since the analysis of account-based systems was limited to EOAs.
- Covering transaction types excluded here: Taproot for Bitcoin and Litecoin, Litecoin's Mimblewimble Extension Blocks (MWEB), and multisignature active-key identification for t-of-n P2MS/P2SH spends.
- Applying the transferred-clustering technique more broadly as a general tool for bringing UTXO heuristic clusters into account-based systems, and using key-based links to benchmark existing heuristics.
Target Audience
Researchers and practitioners in blockchain privacy, forensics and cryptoasset analytics; wallet and protocol designers concerned with key derivation standards; compliance and investigation teams at exchanges and analytics firms; and graduate students in security/privacy who want a concrete example of how cryptographic key-level analysis outperforms address-level heuristics.
Authors’ abstract
It is well known that reusing cryptocurrency addresses undermines privacy. This also applies if the same addresses are used in different cryptocurrencies. Nevertheless, cross-chain address reuse appears to be a recurring phenomenon, especially in EVM-based designs. Previous works performed either direct address matching, or basic format conversion, to identify such cases. However, seemingly incompatible address formats e.g., in Bitcoin and Ethereum, can also be derived from…
The full abstract on arXiv