"They sent it through Tornado Cash." These six words strike fear into every crypto recovery professional. When stolen cryptocurrency enters a mixer, the direct on-chain link between the source and destination addresses is broken. The blockchain still records every transaction, but the connection between the inbound and outbound transactions is obscured — the funds are pooled with other users' funds and redistributed, making it impossible (or so it seems) to determine which outbound transaction corresponds to which inbound transaction. For years, mixers were considered the ultimate dead end in crypto tracing.

But the landscape has changed. The OFAC sanctioning of Tornado Cash in August 2022, the seizure of ChipMixer in March 2023, and advances in blockchain forensics have shifted the balance. Mixers are no longer the impenetrable barrier they once were — though they remain a significant challenge. In this article, I explain how mixers work, why they break the tracing trail, what forensic methods can sometimes see through the fog, and what recovery options exist when funds have been mixed.

How cryptocurrency mixers work

A cryptocurrency mixer (also called a tumbler or blender) is a service that obscures the connection between the sender and recipient of cryptocurrency. The mixer takes deposits from multiple users, pools them together, and sends out equivalent amounts (minus fees) to the users' specified destination addresses. Because the output funds come from the pool — not from any specific input — there is no direct link between any individual input and output.

There are two main types of mixers:

Centralized mixers

Centralized mixers are operated by a single entity. The user sends cryptocurrency to the mixer's address and specifies a destination address. The mixer holds the funds for a period (to break the timing correlation) and then sends equivalent funds (minus a fee, typically 1-3%) to the destination address. The mixer knows the connection between the input and output — it is recorded in the mixer's internal database. This means the mixer itself is a single point of failure: if the mixer's database is seized (as happened with ChipMixer), all the connections are exposed.

Centralized mixers are also vulnerable to theft — the operator can simply steal the deposited funds and disappear. This has happened multiple times (e.g., the Bitcoin Fog case, where the operator was convicted by the DOJ in 2023). Users of centralized mixers must trust the operator not to steal and not to cooperate with law enforcement — a fragile trust.

Decentralized mixers

Decentralized mixers operate through smart contracts, without a central operator. Tornado Cash is the most well-known example. The user deposits cryptocurrency into the Tornado Cash smart contract and receives a cryptographic note (a secret). Later, the user (or anyone with the note) can withdraw the same amount from the smart contract to a different address. The connection between the deposit and the withdrawal is protected by zero-knowledge proofs — the smart contract verifies that the withdrawer has a valid note, but it does not reveal which deposit the note corresponds to.

The key difference from centralized mixers is that there is no central database of input-output connections. The connection exists only in the user's possession (the cryptographic note). If the user loses the note, even they cannot prove which deposit corresponds to which withdrawal. For forensic purposes, this means there is no database to seize — the mixer's records exist only on the blockchain, in the form of encrypted proofs that cannot be decrypted.

Tornado Cash: how it works technically

Tornado Cash operates on Ethereum (and several other EVM-compatible chains). It uses a Merkle tree structure to record deposits without revealing the connection to withdrawals:

  1. Deposit: The user generates a random secret and a nullifier. They compute a commitment (a hash of the secret and nullifier) and send it, along with the cryptocurrency, to the Tornado Cash smart contract. The smart contract adds the commitment to the Merkle tree. At this point, the user's address is publicly linked to the deposit.
  2. Wait: The user waits for a period of time — ideally, long enough that many other users have deposited funds into the same pool. This makes timing-based correlation harder.
  3. Withdrawal: The user generates a zero-knowledge proof that they know a secret corresponding to a commitment in the Merkle tree, without revealing which commitment. They submit this proof and a destination address to the smart contract. The smart contract verifies the proof, checks that the nullifier has not been used before (to prevent double-spending), and sends the cryptocurrency to the destination address. The destination address has no visible link to the deposit address.

The zero-knowledge proof is the key innovation. It proves that the withdrawer is authorized (they know a valid secret) without revealing which deposit they are withdrawing from. An observer (including a forensic analyst) can see all deposits and all withdrawals, but they cannot link any specific deposit to any specific withdrawal.

Tornado Cash pools are organized by denomination — there are pools for 0.1 ETH, 1 ETH, 10 ETH, 100 ETH, and corresponding pools for USDC, DAI, and other tokens. The larger the pool (more deposits), the better the anonymity — because there are more possible sources for any given withdrawal. The 10 ETH pool typically has thousands of deposits, making statistical correlation very difficult.

Forensic methods for tracing through mixers

Despite the cryptographic protections, mixers are not completely opaque to forensic analysis. Several methods can sometimes see through the fog:

1. Timing analysis

If a user deposits funds into a mixer and withdraws them shortly after, the timing correlation can link the deposit and withdrawal. For example, if a scammer deposits 10 ETH into Tornado Cash and withdraws 10 ETH from a fresh address 30 minutes later, the analyst can infer (with some probability) that the withdrawal is from the same user. This method works best when the mixer has low volume (few other deposits to confuse the analysis) and when the user withdraws quickly.

Professional forensic tools (Chainalysis, TRM Labs) use probabilistic timing analysis to assign confidence scores to deposit-withdrawal pairs. The analysis considers the time gap, the denomination match, the number of other deposits in the pool, and other factors. The result is not a definitive link but a probability — "there is an 85% chance that this withdrawal corresponds to that deposit." For law enforcement purposes, a high probability may be sufficient to obtain a warrant or freeze request.

2. Address clustering after withdrawal

Even if the mixer obscures the link between the deposit and withdrawal, the funds eventually reach an exchange or another service where the user's identity is known. If the withdrawn funds are sent to an exchange, the exchange knows the account holder's identity. The forensic analyst can work backward from the exchange deposit to the Tornado Cash withdrawal, and then use timing analysis to link the withdrawal to the original deposit.

This method is particularly effective when the mixer user makes operational mistakes — using the same address for deposit and withdrawal, withdrawing to an address that has been used before, or withdrawing in unusual denominations. In practice, many mixer users (particularly criminals who are not sophisticated) make these mistakes.

3. Relayer analysis

Tornado Cash uses "relayers" — third parties who submit the withdrawal transaction on behalf of the user, in exchange for a fee. The relayer pays the gas cost for the withdrawal transaction and deducts a fee from the withdrawn funds. The use of a relayer means the withdrawal transaction is sent from the relayer's address, not the user's address. However, the relayer knows the destination address (because they must send the withdrawn funds there). If the relayer's records can be obtained (through a legal request or a data breach), the connection between deposits and withdrawals can be reconstructed.

Several Tornado Cash relayers have been identified, and their on-chain activity has been analyzed. In some cases, relayers have been served with legal requests to disclose their records. This has led to the identification of mixer users — including, reportedly, the North Korean Lazarus Group, which used Tornado Cash to launder over $455 million in stolen cryptocurrency.

4. Chainalysis and TRM Labs proprietary methods

Both Chainalysis and TRM Labs have developed proprietary methods for tracing through mixers. These methods are not publicly disclosed (for obvious competitive and operational reasons), but they are believed to combine timing analysis, address clustering, behavioral profiling, and machine learning. According to public statements by both companies, their mixer tracing capabilities have improved significantly since 2022, and they can now trace through some Tornado Cash transactions with moderate confidence.

In one notable case, Chainalysis assisted the FBI in tracing cryptocurrency stolen by the Lazarus Group through Tornado Cash. The tracing led to the identification of $30 million in laundered funds at a cryptocurrency exchange, which was subsequently seized. This case demonstrated that mixer tracing, while not perfect, can produce actionable results when performed by skilled analysts with the right tools.

The OFAC sanctions effect

In August 2022, the US Treasury's Office of Foreign Assets Control (OFAC) sanctioned Tornado Cash — adding the mixer's smart contract addresses to the SDN (Specially Designated Nationals) list. This was unprecedented: it was the first time OFAC sanctioned a piece of software (rather than a person or entity). The sanctions made it illegal for any US person to interact with Tornado Cash.

The sanctions had an immediate effect on Tornado Cash usage. Transaction volume dropped by over 80% within days, as compliant users (and many non-compliant ones) stopped using the mixer. Many DeFi protocols (Uniswap, Aave, dYdX) blocked addresses that had interacted with Tornado Cash, preventing them from using their platforms.

For tracing purposes, the sanctions had two significant effects:

However, the sanctions also led to the emergence of alternative mixers. Several "Tornado Cash forks" appeared on other blockchains (including on Ethereum layer-2s and on non-EVM chains). These alternatives do not have the same level of scrutiny, and some scammers have migrated to them. The cat-and-mouse game between mixers and tracers continues.

Bitcoin mixers: ChipMixer and Bitcoin Fog

On the Bitcoin side, the two most prominent mixers — ChipMixer and Bitcoin Fog — have both been shut down by law enforcement. ChipMixer was seized by US and German authorities in March 2023, and its operator was charged with money laundering. Bitcoin Fog's operator was convicted by a US federal jury in 2023.

The seizure of ChipMixer was particularly significant because the mixer's internal database was obtained by law enforcement. This database contained the connections between deposits and withdrawals — effectively unmixing all the transactions that had ever gone through the service. Law enforcement was able to identify users of ChipMixer, including ransomware groups and other criminals. Several seizures of mixed funds followed the ChipMixer takedown.

The lesson from ChipMixer and Bitcoin Fog is that centralized mixers are inherently vulnerable — their operators can be identified, their infrastructure can be seized, and their records can be obtained. Decentralized mixers (like Tornado Cash) are harder to shut down because there is no central operator to arrest and no central database to seize. But they are not immune — the OFAC sanctions on Tornado Cash showed that the legal system can target the mixer's usage even if it cannot shut down the code.

Can you recover mixed crypto?

The answer depends on several factors:

In our practice, recovery of mixed funds is possible in approximately 20-30% of cases — significantly lower than the recovery rate for non-mixed funds (which is closer to 50-60%). The cases where recovery succeeds typically involve a combination of timing analysis, post-mixer tracing, and law enforcement cooperation. The cases where recovery fails typically involve funds that went through a decentralized mixer (Tornado Cash) to a non-compliant exchange or self-custody wallet.

What to do if your funds have been mixed

If you have traced stolen cryptocurrency to a mixer, do not give up — but set realistic expectations. Here is what you should do:

Case study: partial recovery through Tornado Cash

In a 2024 case, a client lost 85 ETH (approximately $250,000) in a phishing attack. The scammer immediately deposited the ETH into Tornado Cash's 10 ETH pool — eight separate deposits of 10 ETH each, plus one deposit of 5 ETH into the smaller pool. The timing of the deposits (all within a 20-minute window) made it clear that all eight deposits came from the same user.

Our forensic team used timing analysis to identify probable withdrawal addresses. The 10 ETH pool had approximately 2,000 deposits at the time, making direct correlation difficult. However, the scammer made a critical mistake: they withdrew all eight 10 ETH amounts to the same destination address within a 6-hour window. This pattern — eight withdrawals of exactly 10 ETH to the same address in a short time — was unique in the pool's history. Our forensic team was able to identify the withdrawal address with high confidence.

The withdrawal address was traced to a Binance deposit address. We filed a freeze request with Binance, and because the address had received funds from Tornado Cash (a sanctioned entity under OFAC), Binance flagged it automatically. The funds were frozen before the scammer could withdraw them. The client recovered approximately $190,000 (76% of the loss — the reduction was due to the timing of the ETH price and Tornado Cash fees).

This case illustrates several key points: (1) timing analysis can work even through decentralized mixers when the user makes operational mistakes, (2) the OFAC sanctions on Tornado Cash create a secondary freeze mechanism — any address that receives funds from Tornado Cash is tainted and can be frozen by compliant exchanges, and (3) the scammer's mistake (withdrawing all funds to the same address) was the key vulnerability that enabled recovery.

Not all cases end this well. In another case, the scammer withdrew Tornado Cash funds to eight different addresses, each of which sent funds to a different exchange. The funds were dispersed across Binance, Bybit, OKX, and two smaller exchanges. We filed freeze requests with all five exchanges, but only Binance (which had received the largest amount) responded quickly enough. The funds at the other exchanges were withdrawn before the freeze requests were processed. Total recovery was approximately 30% of the loss.

The difference between the two cases was the scammer's operational security. In the first case, the scammer was sloppy — withdrawing to a single address. In the second case, the scammer was careful — dispersing across multiple addresses and exchanges. The lesson is that recovery from mixers is possible but depends heavily on the scammer's mistakes. The more careful the scammer, the harder the recovery.

The bottom line

Mixers remain the single biggest obstacle to crypto fraud recovery. When stolen funds enter a mixer, the direct on-chain trail is broken. But mixers are not impenetrable — timing analysis, post-mixer tracing, law enforcement cooperation, and (in the case of centralized mixers) database seizure can sometimes reconstruct the trail. The OFAC sanctions on Tornado Cash have reduced its usage and made timing analysis more effective, but alternative mixers continue to emerge.

If your stolen crypto has gone through a mixer, recovery is harder but not impossible. Contact us — we can assess the feasibility of tracing through the specific mixer used and develop a recovery strategy that combines forensic analysis, legal tools, and law enforcement coordination.

N. Silinevics
Nils Silinevics Crypto Compliance Counsel · Former FIU Investigator · Valken Legal AG