MARC Record: Structure, Fields and MARC 21

A MARC record is a library catalogue record in a machine-readable format: MARC stands for MAchine-Readable Cataloging, and the version used by most libraries today is MARC 21, maintained by the Library of Congress. Researchers meet MARC records when they export from a library catalogue, import a record into a reference manager, or click a “MARC view” or “staff view” in a catalogue. This guide explains what a MARC record contains, how its parts fit together, and how to read one.

What a MARC record is

According to the Library of Congress introduction to the MARC 21 Bibliographic format, a MARC record consists of three elements: the record structure, the content designation, and the data content. The record structure implements the international standard ISO 2709 (also published as ANSI/NISO Z39.2). The content designation is the set of codes and conventions that identify and characterise each data element. The data content is whatever the cataloguer enters.

The key point for researchers is that MARC is a container, not a set of cataloguing rules. The format does not say how to transcribe a title or what counts as a main author; the cataloguing rules (the Library of Congress introduction names AACR 2 and ISBD as examples) and vocabularies such as the Library of Congress Subject Headings (LCSH) are defined outside the format. Two records for the same book can therefore be valid MARC while differing in the rules used to build them.

Who maintains MARC 21 and what formats exist

MARC 21 is maintained by the Library of Congress Network Development and MARC Standards Office, together with Library and Archives Canada and the British Library. Updates are distributed through the Library of Congress Cataloging Distribution Service and announced on the Library of Congress MARC site. The MARC standards home page lists five formats:

Bibliographic: books, serials, maps, music, computer files and other resources.
Authority: standardised forms of names, subjects and titles.
Holdings: copy and location information for a library’s items.
Classification: classification schedule data.
Community Information: listings of events, organisations and services.

The same site describes MARC development as community-driven, with the Library of Congress as coordinator, through the MARC Advisory Committee, proposals and discussion papers.

MARC 21 versus “MARC”

“MARC” is the general name for the family of formats; “MARC 21” is the harmonised edition in use today. In everyday cataloguing conversation the two are used interchangeably, and a “MARC record” in a current library system is almost always a MARC 21 record.

How to read a MARC record

MARC records are rarely shown in their raw form. Catalogues display them as a table of numbered fields, often under a label such as “MARC view” or “staff view”. The anatomy below comes from the Library of Congress introduction.

Leader. The first field is the leader, a fixed length of 24 character positions. Among other things it carries the record type in position 06 (for example, language material, computer file or music).
Directory. Next comes the directory, made up of 12-position entries giving each field’s tag, length and starting location. It ends with a field terminator (ASCII 1E). This is why a raw MARC record is not a plain-text table.
Tags. Each variable field has a three-character numeric tag. The first digit groups fields by function: 0XX control and identification numbers, 1XX main entries, 2XX titles and imprint, 3XX physical description, 4XX series, 5XX notes, 6XX subject access, 7XX added entries and linking, 8XX series added entries and holdings, and 9XX local use.
Control fields versus data fields. Control fields (tags 00X) have no indicators or subfields. Data fields have two indicator positions (blank is shown as #) and one or more subfields.
Subfields. Each subfield is a delimiter (ASCII 1F, displayed as $ or a similar symbol) followed by a one-character code.
Fill and “unknown” characters. A vertical bar (|) is the fill character, the code “u” generally means the value could not be determined, and “n” means it does not apply.
Record end. The last field is followed by a record terminator (ASCII 1D).

Two fields worked through

The Library of Congress documentation for field 245 (Title Statement; page dated December 2025) gives this display example:

245 10 $a Welcome aboard! : $b your career as a flight attendant / $c Becky S. Bock, Cheryl A. Cage.

Reading it: tag 245; first indicator 1 means a title added entry is made; second indicator 0 means no nonfiling characters (a second indicator of 4, for example, would tell a system to ignore the first four characters, such as “The “, when sorting); $a is the title proper, $b the remainder of the title, and $c the statement of responsibility.

For subjects, field 650 (Subject Added Entry, Topical Term) with second indicator 0 signals that the heading comes from LCSH. The Library of Congress example is:

650 #0 $a Amish.

Strengths and limits

Strengths	Limits
A shared exchange format: libraries can copy and share records because tags, indicators and subfields mean the same thing across systems.	The record is a container only; descriptive quality depends on the cataloguing rules and vocabularies applied outside the format.
Fine-grained coding: indicators and subfields let systems filter, sort and index parts of a field (for example, ignoring leading articles when filing titles).	The raw ISO 2709 structure, with its leader, directory and control characters, is not human-readable or easy to edit without software.
Converters exist: the Library of Congress publishes tools and stylesheets for MARCXML and conversion specifications for BIBFRAME.	The format was designed for library catalogues; the Library of Congress describes BIBFRAME as a Linked Data approach intended to provide a path beyond MARC 21.

Alternatives and related formats

Format	What it is (per its maintainer)	Relation to MARC
MARCXML	A Library of Congress framework for working with MARC data in an XML environment, with schemas, stylesheets and tools.	Represents MARC data in XML; the Library of Congress provides converters between ISO 2709 MARC and MARCXML.
MODS	A Library of Congress schema listed alongside MARCXML and MADS under “MARC in XML”.	Can carry selected data from MARC 21 records.
BIBFRAME	A Linked Data based Library of Congress initiative.	Aims at a path beyond MARC 21 while keeping MARC’s exchange benefits; the Library of Congress publishes MARC 21 to BIBFRAME conversion specifications in both directions.
Dublin Core	A separate metadata vocabulary, not part of the MARC family.	Not a MARC format; see the CASRAI dictionary entry.

Book-trade metadata uses a different standard again: see ONIX. For how the Library of Congress classification numbers that appear in MARC call number fields are built, see Library of Congress classification numbers. More generally, a MARC record is one kind of metadata.

FAQ

What is MARC 21?

MARC 21 is the current harmonised edition of the MARC (MAchine-Readable Cataloging) formats, maintained by the Library of Congress with Library and Archives Canada and the British Library. It defines the structure and coding of bibliographic, authority, holdings, classification and community information records.

What is the difference between MARC and MARCXML?

MARC (ISO 2709) is the compact record structure with a leader, directory and control characters. MARCXML is the Library of Congress framework for representing the same MARC data in XML, with converters in both directions.

Is a MARC record the same as a citation?

No. A MARC record is a cataloguing record built from tagged, coded fields, as described above. A citation is a formatted reference in a citation style, which is a different structure.