Using Trezor in Air-Gapped Mode: Building a Fully Offline Signing Setup Without Internet Dependency

A cryptocurrency holder with significant assets faces a practical security problem: keeping private keys offline while still being able to sign and broadcast transactions. Connecting a hardware wallet to any internet-connected computer introduces a potential attack surface, even with device isolation. An air-gapped setup—where the signing device never connects directly to the internet—removes that surface entirely. The tradeoff is operational complexity: transactions must be constructed on an online computer, transferred to an offline signing device, signed securely, and then broadcast back to the blockchain without the offline device ever touching the network.

Trezor hardware wallets are designed for exactly this workflow. The device signs transactions internally, keeping private keys permanently offline and never exposing them to external software or networks. An air-gapped configuration takes that isolation further by ensuring the signing device itself has no internet capability whatsoever. This requires deliberate setup, careful procedural discipline, and understanding what each component can and cannot protect. The result is a custody model where a theft of the online computer, a network compromise, or a software vulnerability on the signing device’s paired system cannot move funds without physical access to the hardware wallet and knowledge of its PIN and passphrase.

The architecture of offline signing with Trezor

An air-gapped Trezor setup requires two separate systems. The first is an internet-connected computer—the “online” machine—where blockchain data is monitored, transaction details are constructed, and signed transactions are broadcast to the network. The second is an isolated computer or device—the “offline” machine—where the Trezor hardware wallet connects exclusively via USB, never touching any network. The offline machine runs software that can receive unsigned transaction data, display it to the user, and send the signed result back to the online system without ever connecting to the internet.

The key separation point is transaction signing. The Trezor device itself is stateless regarding what it is asked to sign. It does not maintain a balance, verify that an address belongs to your wallet, or confirm that the recipient is legitimate. The device simply takes the transaction data presented to it, displays key details on its screen for user verification, and signs internally using the private key if the user confirms. This design means the transaction construction step—determining amounts, addresses, and fees—happens on the online computer, but the actual signing, which could expose private keys, happens on the offline device without any network connection.

Transaction data is transferred between systems using one of two methods: USB-only transport or QR codes. USB-only transport means the Trezor device itself is the only channel through which information moves between the online and offline computers, physically carried back and forth. QR codes allow unsigned transaction data to be encoded and displayed on the online computer, scanned by a camera connected to the offline machine, signed by Trezor, and then displayed as a QR code on the offline machine to be scanned back into the online system. Neither method requires the offline machine to have internet access, Bluetooth, or any network capability.

Setting up the offline computer and its software

The offline machine should ideally be a dedicated device, not a personal computer used for other purposes. A used laptop running an open-source operating system such as Linux, with network hardware disabled or removed, provides a strong foundation. Some users use a Raspberry Pi or similar single-board computer, though the smaller screen and keyboard options may make verification of transaction details more cumbersome. The critical requirement is that the machine has never connected to the internet and will not connect to the internet; wireless modules should be physically removed if possible, and ethernet should never be plugged in.

Operating system selection matters. Many users choose a minimal Linux distribution with a small attack surface and no bloat. Tails OS, an amnesic live operating system designed for anonymity and offline security, is a popular option because it leaves no persistent storage; every boot is clean and previous sessions are erased. Other choices include a standard Linux distribution installed to a dedicated partition or USB drive, updated once and then never touched again. The principle is to minimize the amount of code that could possibly execute on the offline machine and limit the scope of what needs to be kept secure.

Software compatibility is more restrictive on the offline machine than on a standard setup. Trezor Suite, the official desktop application, includes features designed for internet-connected use, such as real-time price feeds and automatic blockchain synchronization. For air-gapped operation, a simpler tool may be more appropriate. Some users use Electrum (for Bitcoin), which has a built-in watch-only mode and QR code signing features. Others use lighter tools or custom scripts specifically designed for offline transaction construction. The choice depends on which cryptocurrencies need to be managed and whether the offline system will include key derivation, or whether keys are imported from the Trezor device itself.

Constructing transactions on the online system

The online computer is where blockchain monitoring and transaction assembly happen. This system can run Trezor Suite in watch-only mode, meaning it has the Trezor’s extended public keys—data that allows address derivation and balance tracking—but not the private keys themselves. It can see your complete transaction history, current balance across all addresses, and fees on the current network. This information is used to construct transactions: selecting which previous outputs to spend, specifying the recipient address, setting the amount, and calculating appropriate fees.

Once transaction details are assembled, the unsigned transaction data is exported in a format that the offline system can import. This is typically a raw transaction file or a QR code, depending on the software being used. The data contains the transaction inputs (which previous outputs are being spent), outputs (recipient address and amount), and other metadata, but it does not contain any signatures. At this stage, the transaction is technically incomplete and would be rejected by the network if broadcast.

The online computer must have accurate blockchain data to construct valid transactions. Running a full node on the online system ensures that you are not trusting external services to provide accurate unspent transaction output (UTXO) data. However, a full node uses significant storage and bandwidth, and synchronization can take time. An alternative is to use a trusted lightweight node or a personal node accessed only over your local network. The tradeoff between self-sufficiency and convenience is a choice each user must make based on their threat model and technical capacity.

Signing transactions on the offline device

The unsigned transaction is transferred to the offline computer, either by USB-connected Trezor or by QR code, depending on the method chosen. On the offline system, the transaction data is imported into the wallet software. The Trezor device is then connected to the offline computer via USB. The software displays the transaction details on screen: the receiving address, the amount, the network fee, and the total amount being sent. This is the critical moment for user verification. The user should carefully check the address against a known safe source—not anything that could have been modified by malware, but an address recorded separately before the transaction was initiated.

If the details are correct, the user enters their PIN on the Trezor device itself. The PIN is always entered on the physical device, never typed into the computer, because the computer might be compromised. After the PIN is confirmed, the user may be prompted for a passphrase if one was configured. Passphrases are an optional additional layer: unlike the PIN, which is fixed, a passphrase can change the set of addresses derived from the Trezor’s seed, allowing different address hierarchies to be created without changing the physical device or recovery seed. Only after both authentication steps does the device sign the transaction internally, using the private key that never leaves the device.

The signed transaction is then exported from the offline computer, again via QR code or USB, and transferred back to the online system. The online computer receives the now-complete, cryptographically signed transaction and broadcasts it to the blockchain network. The Trezor device itself never sends or receives anything from the internet. The offline computer never connects to the internet. Only the online computer touches the network, and it has never had access to the private keys or the unsigned transaction data before it was signed.

Choosing between USB transport and QR code methods

USB-only transport means the Trezor device physically travels between the offline and online computers. The device is connected to the online computer, unsigned transaction data is loaded into a temporary file, the device is disconnected, carried to the offline machine, connected there, the transaction is signed, the device is disconnected again, and returned to the online machine for broadcast. This method requires the Trezor device itself to be the only data channel. It is conceptually simple and requires no additional hardware on either machine. The weakness is operational friction: each transaction involves multiple physical connections and file transfers, which can be error-prone if not carefully documented.

QR code methods eliminate the need to move the device. Instead, the online computer displays a QR code encoding the unsigned transaction. A camera connected to the offline machine (or even a smartphone used as an offline device) scans the code, the transaction details are displayed on the offline system, the user confirms them, the Trezor signs, and the signed transaction is displayed as another QR code. The online computer’s camera scans this result, and the transaction is broadcast. This reduces physical device movement but introduces a camera as a data channel. Both the online and offline systems must have camera hardware, or a smartphone must be used as the intermediate device for scanning.

The security difference between these methods is subtle. USB transport means all data flows through the Trezor device, which has no ability to be compromised in a way that would modify transaction data in transit. QR codes are unencrypted images and could theoretically be intercepted and modified if someone had physical access to the camera feeds. In practice, the user verifies the transaction details on the offline device’s screen before signing, so modification would require altering that displayed data as well, which is much more difficult. Both methods are substantially more secure than any internet-connected signing approach.

Recovery and backup considerations for air-gapped setups

The Trezor device comes with a recovery seed—a set of words that can restore the private keys if the device is lost or damaged. This seed must be backed up and stored securely, completely separately from both the online and offline computers. The seed should be written down and stored in a physically secure location such as a safe deposit box, a home safe, or multiple geographic locations. Digital backups of the seed should not exist; if a digital backup is made, it should be encrypted with a strong passphrase and stored offline, not on any machine connected to the internet.

The Trezor ecosystem includes recovery options, but recovery itself introduces risk. If the seed must be entered into any device to restore the wallet, that device could be compromised. Some users mitigate this by writing the seed in a form that requires assembly or decoding, or by splitting the seed across multiple locations so that no single stolen backup would expose the complete key material. These are advanced techniques and should only be attempted by users comfortable with cryptography and with multiple test recoveries already completed in a safe environment.

The offline machine itself should also have a tested recovery plan. If the offline computer fails or is lost, can you recover your ability to sign transactions? If the offline machine uses Tails OS or a live system, recovery is simple: any machine with the same OS can be booted from a USB drive and transaction signing can continue. If the offline system is a dedicated laptop with a custom configuration, recovery requires either a backup of the entire disk or detailed documentation of the software setup. Testing recovery before it is needed is critical. A single failure at the moment you actually need to move funds could result in access problems during a time-sensitive situation.

Practical attack scenarios and what air-gapping protects against

An air-gapped setup with Trezor protects against several threat categories. If the online computer is compromised by malware, the attacker can see transaction history and balance, but cannot sign transactions or move funds. Stealing the Trezor device requires both the PIN and optionally a passphrase to access the private keys; without these, the device itself is useless. A network-level attack, such as a compromised internet service provider or a man-in-the-middle intercept, cannot reach the signing device because it has no network interface. Software vulnerabilities in wallet applications running on the offline machine cannot steal private keys because the device performs signing internally.

The setup does not protect against every threat. If someone has physical access to both the Trezor device and the PIN, they can steal the funds. Brute-force PIN attempts are rate-limited by the device with increasing delays, but sufficient physical access over time could eventually crack the PIN, especially if it is a weak pattern. A passphrase adds protection, but a weak passphrase or one derived from personal information could be guessed. Malware on the offline machine could potentially display fake transaction details to the user, causing them to sign a transaction they did not intend. This is why careful address verification and understanding exactly what is being signed remains important.

The online computer could theoretically be compromised in ways that cause you to broadcast transactions that were not intended, though this would not directly steal funds. For example, malware could alter displayed balances or fees to convince you to sign a much larger transaction than you thought. The protection is that the transaction data is verified by the offline machine’s display before signing. The risks are real, which is why self-custody with hardware wallets requires ongoing attention to security practices, not just one-time setup.

Operational security practices for sustained air-gapped operation

Running an air-gapped setup is not a set-and-forget configuration. Sustained security requires consistent practices every time a transaction is constructed and signed. First, always verify the receiving address against a source that could not have been compromised. Write down critical addresses before any transaction; do not copy and paste. Second, confirm the amount being sent matches your intent exactly, including the decimal places. A common scam involves malware altering the amount between what you intended and what you confirm, especially with cryptocurrencies that have many decimal places.

Third, do not rush through PIN or passphrase entry. If you enter your PIN incorrectly on purpose to check your memory before committing, the device will register failed attempts and slow down subsequent tries. Keep detailed documentation of which transaction corresponds to which blockchain transaction ID, so that you can verify broadcast success. If a broadcast fails, do not immediately repeat the transaction; wait a reasonable time and check whether the original transaction appeared on the blockchain. Double-spending the same inputs accidentally is possible if both transactions are broadcast, resulting in one being rejected and one being confirmed at random.

Update strategy is also important. The offline machine should have security updates applied, but not automatically. A tested, deliberate process is safer than automatic updates that could introduce unknown changes. For the Trezor device itself, firmware updates can be performed on the online machine and then installed onto the device when it is next connected. Updates fix security vulnerabilities, so they should be applied, but again, testing and understanding what the update contains is preferable to blind automatic updating. For the online machine, security updates are essential, but they should be applied when you are not actively managing transactions, and critical transactions should be delayed until the update process is complete and stable.

When air-gapping is worth the complexity

Air-gapped operation adds operational friction for every transaction. Moving the device between systems, scanning QR codes, verifying data on two separate displays, or constructing transactions on a less convenient interface all require more time and attention than using a connected hardware wallet. The security benefit is substantial, but it is not unlimited. Air-gapping protects against network-level attacks and compromises of the primary computing environment, but it requires that the user reliably performs the same security checks every time and protects the offline device’s operating system from compromise.

Air-gapping is most justified for funds that are held long-term with infrequent movement. If you need to transact multiple times per week, the operational cost of air-gapping probably outweighs the benefit, and standard hardware wallet operation with a clean, updated, security-conscious online machine may be more practical. If your cryptocurrency holdings represent life-changing amounts or are held in high-threat environments, air-gapping becomes much more attractive. The decision is personal and depends on the amount being secured, how often it moves, the user’s technical comfort, and the specific threats they face.

For maximum security without air-gapping, running a hardware wallet on a rarely-used, dedicated computer that is updated infrequently and never used for other purposes is a strong alternative. This eliminates daily exposure to network compromises and malware while avoiding the operational complexity of air-gapping. The best security setup is one that the user will actually follow consistently, not the theoretically most secure setup that is abandoned as too inconvenient. Air-gapping should be adopted only if the user commits to understanding the procedure, testing it thoroughly before managing significant funds, and maintaining the discipline required for sustained secure operation.