Self-custody concentrates risk in you: your key, your memory, your continued availability. Qualified custody moves most of that to an institution and hands you counterparty risk instead. For significant wealth the deciding factor is rarely security. It is that a custodian has a process for death, incapacity, and disputed authority, and a hardware wallet in a safe does not.
The short version
- Both are defensible. The trade is key risk against counterparty risk.
- Self-custody fails most often through access loss, not theft: a forgotten passphrase, an unrehearsed recovery, a sole holder who dies.
- Qualified custody fails through the institution: insolvency, internal controls, account terms. FDIC and SIPC do not cover crypto.
- Custodians give you something self-custody structurally cannot: third-party evidence that the account is yours, and a documented process when you cannot act.
- Most families holding significant amounts end up splitting by tier rather than choosing one.
What each model actually risks
Self-custody. You hold the key, so nobody can freeze, seize by process, or lose your assets through their own insolvency. The corresponding exposure is total and concentrated. A lost seed, a passphrase held only in memory, a device nobody else can locate, or a death without a rehearsed recovery path all produce the same outcome, and it is unrecoverable.
Qualified custody. The institution holds keys under its own controls, with a defined process for authority changes, death, and incapacity. The exposure moves to solvency, internal controls, and account terms. Neither FDIC nor SIPC covers digital assets, so whatever protection exists is commercial and its limits, exclusions, and named insured decide whether it reaches you.
Worth stating plainly: neither model is safer in the abstract. They fail differently, and one failure mode may be far more likely in your circumstances than the other.
The factor that usually decides it
Not security. Continuity.
A custodian has a procedure for a death certificate, for an incapacity, for a change of authorized signer, for a dispute between members of an entity. Those procedures exist because institutions have handled them for a century. Self-custody has whatever you wrote down and rehearsed.
For a single holder with modest amounts and no dependents, that gap may not matter. As soon as somebody else depends on those assets being reachable, it becomes the dominant consideration, and it is the reason most significant holdings end up at least partly custodied.
The second factor is evidence. A custodial account opened after identity checks by a regulated third party is the strongest available proof that the assets belong to whoever the account names. Self-custody produces only what you wrote, unverified by anyone.
Where the regulatory label fits
“Qualified custodian” comes from the SEC’s custody rule and describes charter and registration status: a bank or savings association, a registered broker-dealer, a registered futures commission merchant, or a qualifying foreign financial institution. It says nothing about key management, insurance, or competence, and the rule binds registered investment advisers rather than you.
Treat it as a threshold that narrows the field, then do the operational diligence separately. What “qualified” actually means covers the detail.
What I actually see
The debate gets conducted as an ideology and decided by circumstances. People who hold strong views about self-custody move to a custodian when a spouse needs access, or when a business partner does, or after a scare with a device.
The genuinely common failure in self-custody has nothing to do with attackers. It is a person who set everything up correctly, told nobody the details, never rehearsed the recovery, and assumed the arrangement was legible to someone else. It usually is not.
On the custodial side, the failure is treating the choice as finished once made. Custodians are counterparties and deserve periodic review: the account terms, the entity holding your account, the reconciliation practice, the concentration. A family that would never hold a single stock will hold a single custodian without noticing they made a comparable decision.
The arrangement I would push most people toward is boring: a custodian for the bulk, self-custody for a working balance, both documented, and the recovery path rehearsed once a year.
Where this goes wrong
Choosing one model for everything and never revisiting it.
The specific failures on each side. Self-custody: recovery material in one person’s memory, no rehearsal, no inventory, and an estate that cannot know what to look for. Custody: assumed federal insurance that does not exist, an account with an affiliate rather than the chartered entity, terms that disclaim the relationship the customer assumed, and the whole position with one provider.
The shared failure is worse than either: assets self-custodied while the paperwork describes an entity or a trust, so neither the key nor the record supports the arrangement everyone believes is in place.
The decision rule
Split by tier and decide each on continuity.
- Anything others depend on reaching belongs where a documented process exists. Usually a custodian.
- A working balance can stay self-custodied, sized so its loss is survivable.
- Rehearse the self-custodied path with whoever would use it, using a real transaction.
- Diligence the custodian operationally, not just on charter: who reconciles, how often, who checks.
- Do not concentrate the custodied portion with one provider.
- Review annually, because people, devices, and providers all change.
If you cannot say today who reaches each tier when you cannot, that is the work, whichever model you prefer.
Where this sits
This is the top-level custody choice. What custody is covers the two jobs underneath it. What “qualified” means narrows the field. What happens if a custodian fails is the risk you accept by moving. Succession is the problem the whole choice is really about.
Custody debates are usually framed as security. For anyone with dependents, the honest framing is continuity.
Sources
- 17 CFR 275.206(4)-2, Custody of funds or securities of clients by investment advisers
- FDIC, Deposit insurance
- SIPC, What SIPC protects
- Wyoming digital asset statutes, Wyo. Stat. Ann. §§ 34-29-101 to 34-29-102
- NIST, Special Publication 800-57, Recommendation for key management
- IRS, Digital assets
Related
- What is digital asset custody?
- What is a qualified crypto custodian?
- What happens if a crypto custodian fails?
- How do I move from self-custody to qualified custody?
- Private key succession planning
- Crypto custody
Last updated: 3 August 2026.
This article is general education, not legal, tax, or investment advice. Custody arrangements can reduce certain risks but do not eliminate them, and outcomes depend on your facts, your provider, and your documents. Talk to a qualified attorney about your own situation.
