Most tokenisation projects arrive at a lawyer with the framework already assumed. Usually MiCA, because the token is on a chain and MiCA is the crypto regulation. That assumption is wrong roughly as often as it is right, and getting it wrong is not a detail. It decides whether you need a prospectus, whether you need a MiFID licence, and whether the thing you have been calling a utility token is in fact a transferable security.
This is the decision path, in the order it actually has to be taken.
The first question is the only one that matters
Is the token a financial instrument under MiFID II?
Everything follows from this, because MiCA explicitly excludes crypto-assets that qualify as financial instruments. The two regimes do not overlap and they do not compete. One of them applies.
If the token represents a share, a bond, a derivative or units in a fund, it is very likely a financial instrument, and wrapping it in a smart contract changes nothing about that. Tokenisation is a change in how an asset is recorded and transferred. It is not a change in what the asset is. This is the principle that catches teams who assumed the technology created a new legal category.
If it is a financial instrument
You are in the traditional regime, and the whole of it. MiFID II governs the instrument and the services around it. The Prospectus Regulation governs the offer. CSDR governs settlement and central registration. MAR governs market abuse.
None of these were written with a ledger in mind, which is the real friction. CSDR in particular assumes a central securities depository, and a token whose transfer is its settlement does not sit comfortably with that assumption.
That friction is what the DLT Pilot Regime exists to relieve. It lets market infrastructures trade and settle tokenised shares, bonds and UCITS on a ledger under targeted exemptions from MiFID II and CSDR, while keeping investor protection and market integrity intact. Note what it is and is not: it is a regime for infrastructure operators, not a licence for issuers. If you are issuing rather than operating a venue, the Pilot Regime is context, not your answer.
If it is not a financial instrument
Now you are in MiCA, and the next question is which of three categories.
Asset-referenced token (ART) if it references multiple currencies, assets or a basket to maintain stable value.
E-money token (EMT) if it references a single official currency.
Other crypto-asset for everything else, which in practice is most utility and access tokens.
The obligations diverge sharply. ARTs and EMTs carry issuer authorisation, reserve, redemption and governance requirements. The third category is comparatively light, principally a white paper with notification and content requirements. Teams naturally want to be in the third category, which is precisely why the classification deserves an adversarial reading rather than an optimistic one.
Where real-world asset tokenisation usually lands
RWA is the term that causes the most confusion, because it describes an economic ambition rather than a legal category.
Tokenised real estate typically works through a special purpose vehicle, and what the holder owns is an interest in that vehicle. An interest in an entity that pools capital and invests it is frequently a fund unit or a transferable security, which means MiFID II and quite possibly AIFMD, not MiCA.
Tokenised debt is debt. Tokenised equity is equity. Tokenised commodities depend on whether the holder has a claim on a physical asset or on a derivative exposure to its price, and those land in different places.
The pattern is consistent: the closer the token gets to a claim on an income stream or on an entity, the more likely it is a financial instrument. Tokens that genuinely stay outside tend to be the ones granting access to something rather than a claim on anything.
The questions to settle before writing any code
In practice the classification turns on a small number of factual questions, and they are worth answering in writing early, because they get harder to answer honestly once the product exists.
- What does the holder actually get? Access to a service, a claim on an entity, a claim on an income stream, or exposure to a price?
- Is it transferable, and is it intended to trade on a secondary market?
- Do holders expect a return, and does that return depend on someone else's efforts?
- Does it reference a currency or a basket to hold its value stable?
- Where are you offering it, and to whom? Retail changes the calculus considerably.
The third question is the one that quietly reclassifies more projects than any other.
Why this is worth settling first
The cost asymmetry is severe. Getting the classification right at the design stage means choosing a structure and building to it. Getting it wrong means discovering after launch that you made an unauthorised offer of transferable securities, which is not a paperwork problem.
National competent authorities have also shown they will look through form to substance. A token described as a utility token in a white paper but functioning as an investment will be assessed as what it does, not as what it was called.
This is the work behind a regulatory audit engagement, and it is deliberately unglamorous. A clear answer to "which regime applies" is the artefact that makes everything downstream, the prospectus question, the licensing question, the venue question, tractable instead of speculative.