Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
ISO 26324:2025 — The Digital Object Identifier (DOI) System
A comprehensive technical guide to the ISO standard defining the syntax, assignment, resolution, and administration of persistent digital identifiers for any object type.
1. Introduction and Scope
ISO 26324:2025 (third edition) specifies the Digital Object Identifier (DOI) system — a social and technical infrastructure for the persistent unique identification of objects of any type. First deployed in 1998 and now widely adopted across academic publishing, cultural heritage, scientific data, and commercial content management, the DOI system provides a framework for creating, registering, resolving, and administering persistent identifiers.
Clarifying “DOI”: The acronym DOI stands for “digital object identifier”, meaning a digital identifier of an object — not an “identifier of a digital object”. This distinction is fundamental: DOI names can identify abstract entities (a novel as a creative work), physical objects (a museum artifact), and non-physical entities (a legal agreement), in addition to digital files and resources.
The scope of ISO 26324:2025 covers the syntax of DOI names, the principles of their assignment and registration, the resolution infrastructure that enables DOI names to be used as actionable network links, and the system metadata framework that describes identified referents. The standard does not prescribe particular implementation technologies.
2. DOI Name Syntax
2.1 General Structure
A DOI name consists of an ordered sequence of Unicode code points (Graphic type, as per ISO/IEC 10646) divided into two parts separated by U+002F Solidus (forward slash): a DOI prefix and a DOI suffix. The DOI prefix itself may contain a directory indicator optionally followed by a registrant code, separated by U+002E Full Stop.
| Component | Example | Description |
|---|---|---|
| Full DOI name | 10.1000/123456 |
Complete identifier: prefix + “/” + suffix |
| Directory indicator | 10 |
Allocated by the Registration Authority; “10” is the only indicator assigned to date (though the 3rd edition now allows others) |
| Registrant code | 1000 |
Unique string allocated to a registrant; may be subdivided (e.g., 1000.11) |
| DOI suffix | 123456 |
Sequence of code points chosen by the registrant, unique within the prefix |
There is no defined limit on the total length of a DOI name, or on the prefix or suffix individually. The suffix can incorporate identifiers from other schemes, such as ISBN, ISSN, ISAN, ISRC, or ISNI, following syntax rules maintained by the Registration Authority.
2.2 Unicode and Case Sensitivity
One of the most technically significant updates in the 2025 third edition is the clarification of case handling. DOI names are case-insensitive only when testing for equivalence and only for the Basic Latin Unicode block (U+0041–U+005A and U+0061–U+007A). This means:
10.5594/SMPTE.ST2067-21.2020and10.5594/sMPTE.sT2067-21.2020are equivalent (same Basic Latin letters, different case)10.26321/Á.GUTIÉRREZ.ZARZA...and10.26321/á.gutiérrez.zarza...are not equivalent (accented characters are outside Basic Latin case mapping)- Pre-composed characters (U+00C1 = Á) and decomposed sequences (U+0041 U+0301 = A + combining acute) are not equivalent even though they render identically
Implementation Warning: When storing or comparing DOI names in databases or application code, use code-point-by-code-point comparison (not normalization). Applying Unicode normalization forms (NFC, NFD) before comparison is explicitly prohibited by the standard. This differs from many other identifier systems and is a common source of bugs in DOI resolution implementations.
2.3 Representation Forms
The standard defines four representation forms for DOI names:
| Form | Example | Use Case |
|---|---|---|
| Visual (plain text) | doi:10.1006/jmbi.1998.2354 |
Print and display; “doi:” prefix optional for clarity |
| URI form | doi:10.1006/jmbi.1998.2354 |
Using the “doi” URI scheme per RFC 3986 |
| URN form | urn:doi:10.1006/jmbi.1998.2354 |
Using the “doi” URN namespace per RFC 8141 |
| HTTP proxy form | https://doi.org/10.1006/jmbi.1998.2354 |
Resolvable URL via the Registration Authority’s proxy service |
Engineering Insight: For modern web applications, the HTTP proxy form (
https://doi.org/...) is the most practical representation, as it provides immediate resolvability in any web browser. When generating DOI links programmatically, always percent-encode the DOI name per RFC 3986 before constructing the URL. The legacy forms https://dx.doi.org/... are deprecated in the 2025 edition.3. Assignment Principles and Persistence
3.1 Core Principles
The assignment of DOI names is governed by several fundamental principles that distinguish the DOI system from simple URL-based identification:
- Uniqueness: Each DOI name specifies one and only one referent within the DOI system
- Opacity: The string itself carries no definitive meaning — no information should be inferred from the sequence of code points
- Persistence: A DOI name shall never change or be deleted, regardless of rights transfers, ownership changes, or even errors in the string
- Granularity: DOI names can be assigned at any desired level of precision — from an entire novel down to a single paragraph, quotation, or image within it
- First-class naming: The identifier has an identity independent of any other object (unlike a URL, which is tied to a network location)
3.2 Granularity in Practice
The standard explicitly recognizes that a registrant may assign DOI names at varying levels of granularity depending on functional need. For example, a publisher might assign separate DOI names to:
- A novel as an abstract work
- A specific edition of that novel
- A specific chapter within that edition
- A table, figure, or even a single quotation
Each of these DOI names is fully independent and resolvable. The granularity choices determine what constitutes a “change” requiring a new DOI name — if a referent is defined to include its ownership, a change of ownership yields a different referent requiring a new DOI name while the original persists.
4. Resolution Mechanisms
The DOI system provides resolution infrastructure that accepts a DOI name as input and returns a resolution record containing current information about the referent. Resolution is independent of the Domain Name System (DNS) but designed to be compatible with it.
The standard specifies the following functional requirements for the resolution service:
| Requirement | Description |
|---|---|
| Internet compatible | Transmission via globally unique address space and communications |
| First-class naming | Identifier existence independent of any other object |
| Functional granularity | Separate resolution for each distinguishable referent |
| Data typing | Extensible definition of constraints on data entries |
| Multiple resolution | Simultaneous return of multiple pieces of current information |
| DNS independent | Not reliant on DNS, but capable of working with it |
| Scalability | No limitations on number of identifiers or string length |
| Unicode compliant | Full support for the Unicode character set |
Engineering Insight: The multiple resolution capability is a powerful but underutilized feature. A single DOI name can resolve to different URLs depending on context — for example, returning a publisher’s landing page for human users, a metadata record for automated harvesters, and a full-text PDF for authorized subscribers. This is achieved through resolution records that contain multiple typed entries rather than a single URL.
5. System Metadata
Each DOI name must be associated with system metadata that describes the referent unambiguously and precisely. The metadata framework is designed to promote interoperability across different applications and communities. The standard defines two categories of metadata elements:
5.1 Reference Elements
These describe the referent itself and include the DOI name, referent type and sub-type, referent name(s), basic metadata sufficient to define what the referent is, and other identifiers referencing the same referent.
5.2 Administrative Elements
These cover the registration authority identifier and the timestamp when the DOI record was created.
The metadata model supports all types of referents through a generic framework — rather than treating books, videos, sound recordings, and photographs as fundamentally different entities with different metadata schemas, they are recognized as creations with different values of the same higher-level attributes.
6. Administration and Registration
The DOI system is administered by the ISO 26324 Registration Authority, which provides the following core services:
- Allocating unique DOI prefixes to registrants
- Maintaining the DOI directory — a single logical directory of all registered DOI names
- Providing resolution infrastructure (the
https://doi.org/service) - Developing and maintaining documentation including the DOI Handbook
- Implementing policies for persistence and interoperability
The Registration Authority may delegate certain tasks to registration agencies (such as Crossref, DataCite, and other DOI registration agencies), which handle prefix allocation and metadata registration for specific communities.
Key Takeaway: ISO 26324:2025 represents the mature evolution of the DOI system — now with clarified Unicode handling, updated representation forms, a streamlined metadata specification, and administrative flexibility through the ability to assign directory indicators beyond “10”. For engineers and architects building persistent identifier infrastructure, the standard provides a robust reference model that balances technical precision with practical implementability.
7. Frequently Asked Questions
Q1: How is a DOI name different from a URL?
A URL identifies a location on the web; if the resource moves, the URL breaks. A DOI name is a persistent identifier that remains constant regardless of where the identified object is located. The DOI resolution infrastructure maintains the current location mapping. Additionally, DOI names can identify non-digital and abstract objects that have no web location at all.
Q2: Can a DOI name contain characters other than ASCII letters and numbers?
Yes. DOI names can contain any Unicode Graphic character, including accented letters, symbols from non-Latin scripts, punctuation, and spaces. However, the standard’s equivalence rules mean that case differences in Basic Latin letters (A–Z) are ignored when comparing DOI names for equivalence, while case differences in other characters (e.g., accented letters) are not ignored. Pre-composed and decomposed character sequences are considered different even if they look the same.
Q3: What happens if a journal changes publisher? Does the DOI change?
No. One of the core principles of the DOI system is persistence — a DOI name shall never change or be deleted, even when management responsibility transfers to a different organization. The new publisher updates the resolution record (typically through the registration agency) so that the existing DOI continues to resolve correctly. This is why DOI names have become the standard for persistent citation links in academic publishing.
Q4: What changed between the 2022 (2nd edition) and 2025 (3rd edition) of ISO 26324?
The three main changes are: (1) clarification that DOI names are case-insensitive only for Basic Latin letters (A–Z/a–z) and only when testing for equivalence; (2) updates to the list of common representation forms (including deprecation of legacy HTTP proxy forms like dx.doi.org); and (3) revision of the system metadata specification to reflect current practice. The previous restriction limiting the directory indicator to “10” has also been removed, allowing for future expansion.
