Weekly MARC dumps of the Library of Congress Subject Headings (LCSH): every subject authority record, downloaded from LC's own id.loc.gov. The records are LC's, unmodified.
468,352 headings. Since 2026-10-14: +149 net (7.1/day), 206 new or changed (9.8/day).
2026-11-07, 468,352 records. Since 2026-11-06: 150 new, 1 changed, 1 deleted.
| lcsh-20261107.mrc.gz | 58 MB | MARC 21 |
| lcsh-20261107.mrcxml.gz | 56 MB | MARCXML |
| manifest-20261107.csv.gz | 8 MB | CSV |
| changes-20261107.mrc.gz | 59 KB | new or changed, MARC 21 |
| changes-20261107.mrcxml.gz | 58 KB | new or changed, MARCXML |
| changes-20261107.csv.gz | 3 KB | new or changed, CSV |
| deleted-20261107.csv.gz | 88 bytes | deleted, CSV |
SHA256SUMS · all files · Atom · all releases on the Internet Archive
| lcsh-20261106.mrc.gz | 58 MB | MARC 21 |
| lcsh-20261106.mrcxml.gz | 56 MB | MARCXML |
| manifest-20261106.csv.gz | 8 MB | CSV |
id.loc.gov, at the pace its robots.txt asks for, updated weekly with whatever LC changed. Only the computed parts of the leader (record length, base address) are recalculated when writing MARC.changes-* hold only the records new or changed since the previous release (compared by content); deleted-* lists the headings removed (LCCN, heading). *-latest names always point to the newest..mrc into readable .mrk. Any system that imports MARC authority records can load it.sh85076432 (no blanks) where LC's catalog has sh 85076432 , so match on
010 $a; and 005 is the date of the last cataloguing change, not LC's August 2025 system migration. LC's internal
fields (035 system numbers, 906/910 staff notes, 999 FOLIO IDs) are not included..gz files of the first two releases were re-packed so they extract with the right name (before, some tools extracted them as .gz.part). The records inside are unchanged; only the checksums changed. If you have the old file, it's fine: just rename it.Thank you to the catalogers and staff of the Library of Congress, who have built and maintained these headings for more than a century and share them freely with libraries everywhere. This mirror exists only because of their work.
Unofficial. Not affiliated with or endorsed by the Library of Congress. Maintained by Alonso Cossío Vázquez (iratedemon04 at gmail dot com).