ERC-20 Token Audit: What Gets Checked and What It Costs
Part of our guide to auditing a smart contract

Quick answer. ERC-20 requires no audit, but a token going public should get an independent review. A token audit checks your custom logic, mint, pause, blacklist and upgrade powers, fee or tax paths, and initialization against the design you intend.
Video transcript
ERC-20 requires no audit. But a public token should get an independent review, and the decision is scope.
A familiar library does not make your token reviewed. Roles, permit, upgrades and transfer taxes each widen scope.
Ask for a power inventory: who holds mint, burn, pause, blacklist, fee setting and upgrade powers, with limits and revocation paths.
Renouncing ownership only disables owner-gated functions. Separate role admins and upgrade authorities can survive it.
Uniswap's version three documentation says fee-on-transfer tokens do not work with its router contracts. Guardian's g8keep review found buy and sell taxes also charged liquidity additions and removals, marked Medium and Acknowledged.
Procur3, an audit marketplace, publishes one thousand to ten thousand dollars for custom ERC-20 scopes. It calls that marketplace data, not a rate card. Guardian publishes no price list and quotes from scope.
Book early. Kraken warns top audit firms run multi-month queues. Review is one phase, followed by your fixes, the auditor's re-review and the final report.
A token audit covers the token and what it inherits. Vesting, claims, staking, lending and bridges must be added or reviewed separately.
A useful report names the scope and commit, each finding's severity, its remediation status, and the risks that remain. Read statuses, not counts.
Read the full guide, or send Guardian your repository, commit and files for a quote.
Does Your Token Need an Audit?
ERC-20 defines an interface: supply, balances, transfers and allowances. It is not a security certification, and it requires no audit.[1]
Kraken's March 2026 pre-TGE playbook lists an audited contract with a public report among what exchanges ask for. It says every High and Critical finding must be resolved and verified by the auditor before launch.[2]
Ethereum's developer guidance recommends an independent review once your own testing is done. It also warns that audits will not find every bug.[3]
So for a public token, the answer is yes. The real decision is scope: what the review must cover for the design you chose.
Listing and legal duties depend on the venue and jurisdiction. For the security questions listing teams raise, see what exchange listing reviews check.
This guide is for issuers buying a review. For the bugs themselves, with code, read our guide to ERC-20 vulnerabilities.
Standard vs Custom Tokens
OpenZeppelin's base ERC20 leaves the supply mechanism to your contract. Its cap, burn and pause extensions each add behavior of their own.[4]
In Contracts 5.x, transfers, mints and burns all pass through one function, _update. Override it, and your rule touches every token movement.
A familiar library does not make your token reviewed. The composition, constructor arguments, roles and deployment settings are yours.
Every review should still confirm the basics. ERC-20 requires a Transfer event on every transfer, zero-value ones included, and an Approval event on each successful approve.[1]
Callers must also handle a false return value. Your token's failure behavior decides how safely others can call it.
Scope grows with each feature you add on top:
| Token design | What the review adds | Scope |
|---|---|---|
| Fixed supply, library code, no admin | Composition, constructor, events, deployment settings | Focused review |
| Capped mint, burn or pause | Who holds each role, and its limits | Add role and admin paths |
| Permit (ERC-2612) | Nonces, deadlines, signatures, domain separation | Add signature handling |
| Upgradeable proxy | Initializers, implementation locking, storage layout | Add the upgrade path |
| Fees, taxes, blacklists or hooks | Every transfer route, exemption and integration | Full custom logic review |
Mint, Pause and Blacklist Powers
Ask the auditor for a power inventory. For each privileged function, it records the holder, its administrator, its limits and how the power is revoked.
The inventory separates two problems: a caller who should not have a power, and an authorized admin who does. OpenZeppelin's access control guide draws the same line.[5]
Role administration is where inventories go stale. An AccessControl admin can grant or revoke its role, and DEFAULT_ADMIN_ROLE administers every role by default.
Renouncing Ownable ownership disables onlyOwner functions only. Separate role admins and upgrade authorities can survive it.
Write the blacklist policy down before the review starts: holders, recipients and approved spenders.
Guardian's raac.io review found a blacklisted spender could still move a holder's tokens through an existing allowance, because only sender and recipient were checked (L-10, Low, Acknowledged).[6]
| Power | What to confirm | Red flag |
|---|---|---|
| Mint | Holder, supply cap, who can add minters | A cap described as ending issuance |
| Burn | Burns take the caller's tokens or an allowance | Burning another holder's balance without allowance |
| Pause | Authorized pause and unpause functions exist | Pausable inherited with no usable pause path |
| Blacklist | Checks match the written policy | Spenders or recipients outside the check |
| Fee setting | Rate bounds and recipient changes | Owner can set any rate |
| Upgrade | Proxy admin holder, initializers, storage layout | Unlocked implementation or unprotected initializer |
Fee and Tax Logic
A transfer tax is custom logic by definition. It changes what recipients receive and what every integration must handle.
Uniswap's v3 documentation says fee-on-transfer tokens do not work with its router contracts.[7]
Name the actual router and version in scope. A token that transfers correctly can still fail the trading path you plan to launch on.
Taxes can also hit flows nobody meant to tax. Guardian's g8keep launchpad review found buy and sell taxes also charged liquidity additions and removals (M-02, Medium, Acknowledged).[8]
Protocols that list your token face the same accounting problem. Our weird ERC-20 tokens checklist shows what their integrators test.
Give the auditor acceptance criteria, not just code. Ask them to test against these:
- Maximum rates, enforced in code, and who can change them.
- Every exemption, and who can add or remove one.
- Rounding direction, and where any remainder goes.
- Fee recipients, and whether an admin can redirect them.
- Direct transfers and delegated transferFrom calls.
- Buys, sells, liquidity additions and removals on your launch pool.
- Transfer events that reconcile with actual balance changes.
Token Audit Cost and Timeline
Price follows scope. Guardian publishes no price list and quotes from the repository, commit and files in scope.[9, 10]
Procur3, an audit marketplace, publishes $1,000 to $10,000 and one to two weeks for custom ERC-20 scopes of about 200 to 500 nSLOC. Its FAQ says most quotes include one fix review.[11]
Procur3 says the range reflects marketplace data, not a rate card, and that quotes can fall outside it. It does not publish the sample behind the token row.
For engineer time, Arbitrum's February 2025 audit program proposal called $10,000 to $40,000 per auditor-week not uncommon, depending on complexity. That is a weekly rate, not a token package price.[12]
Security Alliance gives rough review lengths of one to two weeks under 1,000 lines and two to five weeks from 1,000 to 4,000.[13]
Review is one phase. Security Alliance separates it from your fixes, the auditor's re-review and the final report.
Book early. Kraken's playbook warns that top audit firms run multi-month queues.[2]
Our audit cost estimator and audit timeline guide work through both numbers for any codebase size.
When quotes arrive, compare them on the same terms:
- The same commit and file list sent to every firm.
- Which extensions, deployment scripts and settings are reviewed.
- Which integrations, such as your launch pool and router, are tested.
- Who reviews, and for how many auditor-weeks.
- How many fix reviews are included, and the price of each extra one.
- Every exclusion, in writing.
Token Audit vs Protocol Audit
A token audit covers the token contract and what it inherits. It does not automatically cover vesting, claims, staking, lending or bridges.
Those contracts can hold much of the supply at launch. Add them to scope or buy a separate review; our token launch security checklist covers vesting, claim and liquidity contracts.
Contract reviews also stop at the chain. Guardian's standard smart contract audit excludes offchain coverage, such as signing infrastructure and frontends, and post-launch updates.[9]
An audit is a snapshot of one commit. To decide which later changes need another review, see when you need a re-audit.[13]
Reading the Token Audit Report
A useful report names the reviewed scope and commit, each finding with its severity, its remediation status, and the risks that remain.[9, 13]
Read statuses, not counts. Acknowledged means the report records no fix, so look for the team's stated reason.
An audit badge on a launchpad page is not a report. Ask for the document, the commit and the auditor's own published copy.
Before launch, match the deployed bytecode to the final reviewed commit. Our guide to reading a smart contract audit report walks through every section.
Scope a Review of Your Token
Guardian reviews token contracts at a fixed commit with two independent teams, human-verified findings and a fix review. Send the repository, commit and files in scope for a quote.
Frequently Asked Questions
Is a free token audit enough?
Free automated scans can flag known patterns while you prepare. They do not replace a scoped, independent review that weighs your intended design and verifies each fix. Ethereum's security guide lists code analysis tools as basic checks to run before merging changes.[3]
Should I publish my token audit report?
Yes, once fixes are verified. Kraken's playbook calls a final report published alongside the token contract one of the clearest credibility signals a team can give. Security Alliance notes that web3 audit reports are commonly made public to build community trust.[2, 13]
What changes if my token is upgradeable?
OpenZeppelin warns that incompatible storage changes can corrupt application state, so check each new implementation's storage layout and initializers before the proxy points to it. It also recommends locking implementation contracts so nobody can initialize them directly.[14]
Related reading
- what is a smart contract audit
- token launch security checklist
- erc20 vulnerabilities
- exchange listing security review
Sources
- Ethereum Improvement Proposals: ERC-20: Token Standard
- Kraken: Kraken 360 Pre-TGE Playbook Part 2: The 5-Step Execution Window
- ethereum.org: Smart Contract Security
- OpenZeppelin: ERC20 API Reference
- OpenZeppelin: Access Control
- Guardian: RAAC Core Contracts: L-10, Blacklisted Spender Bypasses RToken Compliance
- Uniswap: Token Integration Issues
- Guardian: g8keep Token Launchpad: M-02, Liquidity Providers Are Charged Trading Fees
- Guardian: Smart Contract Audit Services
- Guardian: Smart Contract Audit Cost Estimator and Calculator
- Procur3: Smart Contract Audit Cost Calculator
- Arbitrum Foundation: Arbitrum Audit Program
- Security Alliance: What to Expect from a Smart Contract Audit
- OpenZeppelin: Writing Upgradeable Contracts
- Ethereum Improvement Proposals: ERC-2612: Permit Extension for EIP-20 Signed Approvals