Military Records for Genealogy Research in the United States

Walk into the National Archives in Washington, D.C., and somewhere in its holdings are pension files thick enough to reconstruct a life — a soldier's physical description, his wife's maiden name, sworn statements from neighbors attesting to his service, sometimes letters in his own handwriting. Military records are among the most information-dense documents in American genealogy, and they reach back further, and cut deeper, than most researchers expect.

• Definition and scope
• Core mechanics or structure
• Causal relationships or drivers
• Classification boundaries
• Tradeoffs and tensions
• Common misconceptions
• Checklist or steps
• Reference table or matrix
• References

Definition and scope

Military records for genealogy purposes encompass the documentary output of a person's relationship with the armed forces — from enlistment through discharge, and often well beyond death. The category spans federal and state holdings, covers conflicts from the colonial militia era through the 20th century's world wars, and includes both the service records themselves and the downstream paper trail that military service generated: pension claims, bounty land warrants, draft registrations, and burial documentation.

The National Archives and Records Administration (NARA) holds the central federal repository for these records, though significant collections also reside at state archives, the Department of Veterans Affairs, and the National Personnel Records Center (NPRC) in St. Louis, Missouri. The scope extends to an estimated 18,000 cubic feet of Civil War pension files alone, according to NARA, a figure that begins to suggest the scale of what's available to researchers working at genealogyauthority.com.

Core mechanics or structure

Military records cluster into four functional types, each generated by a different administrative moment in a soldier's career.

Service records document the fact of service itself — muster rolls, compiled military service records (CMSRs), and descriptive books that captured a soldier's physical appearance, birthplace, and occupation at enlistment. CMSRs are abstracted cards drawn from the original muster rolls and payrolls; they're the fastest entry point but frequently the thinnest in biographical detail.

Pension and bounty land files are the genealogical jackpot. To claim a pension, a veteran or his widow had to prove service, age, marriage, and often the births and deaths of children. These files — particularly those for the Revolutionary War, War of 1812, and Civil War — can run to dozens of pages. The Revolutionary War pension files, digitized through a collaboration between NARA and Fold3, number approximately 80,000 files for veterans and their widows.

Draft registration records span two world wars. The World War I draft registrations — known as the "Old Man's Draft" when the 1942 registrations for men born between 1877 and 1897 are included — cover an estimated 24 million men (NARA, RG 163). World War II draft cards for the 1940 registration cover roughly 16 million men. Both sets include physical descriptions, employer, and next of kin, making them useful for extending family trees into the early 20th century even for men who never deployed.

Burial and memorial records include the Veterans Legacy Memorial database, the General Index to Pension Files, and interment records from 155 national cemeteries administered by the National Cemetery Administration.

Causal relationships or drivers

The volume and richness of a military record is largely a function of two variables: the era of service and the nature of any benefit claimed afterward.

Earlier conflicts generate sparser service records but richer pension files, because pension legislation — particularly the 1818 and 1832 Pension Acts — required veterans to narrate their service in considerable detail. A soldier who filed for a pension under the 1832 Act, which for the first time required a full account of service, left behind material that a muster roll alone could never provide.

The 1973 fire at the NPRC in St. Louis destroyed an estimated 16–18 million Official Military Personnel Files (NARA), covering Army personnel discharged November 1912 through January 1960 and Air Force personnel discharged September 1947 through January 1964. This single event reshaped the research landscape for 20th-century military genealogy more than any policy decision ever could. Reconstruction records — built from Veterans Administration files, pay records, and unit records — exist for a portion of the destroyed files, but gaps remain.

State militia and National Guard records follow a different administrative chain entirely. These pass through state adjutant general offices rather than federal repositories, meaning that researching a state militiaman from the Mexican-American War requires a trip — literal or virtual — to the relevant state archives rather than NARA.

Classification boundaries

Not all military-adjacent documents are military records in the archival sense. Distinguishing between record types prevents wasted research time.

Draft registrations are not service records. A man who registered for the World War I draft but was never called up has a draft card — not a service record, not a discharge document. The two are filed separately and held in different record groups at NARA.

Confederate records occupy a split custody. NARA holds compiled Confederate service records abstracted from captured Confederate records. State-level holdings — particularly Virginia, Georgia, and South Carolina — supplement these significantly. Confederate pension records, however, were administered by the individual Confederate states, not the federal government, and are held at state archives.

Merchant marine and civilian defense worker records are not military records under NARA's classification, despite their holders sometimes receiving veteran status retroactively. The United States Merchant Marine records are held primarily by the U.S. Coast Guard.

For research on African American soldiers — particularly the United States Colored Troops of the Civil War — the pension files are especially significant, as explored further in resources on African American genealogy. The USCT pension files often document family structures that other record types erased entirely.

Tradeoffs and tensions

The two dominant tensions in military records research are access speed versus record completeness, and federal versus state custody.

Online digitization has made CMSRs and draft cards accessible without a trip to Washington or St. Louis. Fold3 and FamilySearch have digitized substantial portions of the Civil War, Revolutionary War, and both world war collections. But digitization has outpaced indexing accuracy — names transcribed from 19th-century handwriting carry error rates that can render a record invisible to keyword search. A soldier named "Johann Schrader" might be indexed as "Schaader" or simply absent from the index while present in the images. Browsing the original images, not just searching the index, remains essential.

The second tension is between the richness of federal records and the completeness of state-level collections. Soldiers who served in state volunteer regiments during the Civil War may have records at the state adjutant general's office that contain details absent from the federal compiled service record. Illinois, for example, maintains regimental records through the Illinois State Archives that supplement the NARA holdings significantly.

Common misconceptions

Misconception: All military records are publicly available.
Records for veterans who served within the past 62 years are restricted under the Privacy Act of 1974 (5 U.S.C. § 552a). Next of kin may request records for deceased veterans; third-party researchers face more limited access. The 62-year default period is not a firm cutoff — specific records may be restricted longer.

Misconception: A pension file means the pension was granted.
Pension claims were frequently rejected, especially early in the pension system's history. A rejected pension application can be more genealogically useful than a granted one, because the claimant often provided extensive supporting testimony attempting to prove eligibility.

Misconception: The 1973 NPRC fire destroyed everything for the affected period.
Reconstruction is possible for a significant portion of destroyed records using alternate sources: VA records, pay vouchers, morning reports, and hospital records. NARA's eVetRecs system provides a starting point for reconstruction requests.

Misconception: Women have no military records.
Women served as nurses through the Army Nurse Corps from 1901 onward, and in the Women's Army Corps, WAVES, and other branches during World War II. Their service records are held at the NPRC and follow the same access rules as male veterans' records.

Checklist or steps

The following sequence reflects the standard research workflow for locating a subject's military records.

1. Search the Access to Archival Databases (AAD) and NARA's catalog for existing digitized holdings in the relevant record group.
2. Check Fold3 and FamilySearch Military Records for digitized and indexed versions before requesting physical reproductions.
3. Cross-reference with vital records, census records, and newspaper archives to corroborate and expand the military record's biographical details.
4. Verify physical descriptions, ages, and birthplaces against other primary sources — primary vs. secondary sources distinctions matter here, since much pension testimony is secondary evidence provided years after the fact.

Reference table or matrix