Device History Record (DHR): Definition and US QMSR
A Device History Record (DHR) connects an identified device, lot or batch to what actually happened in production: instructions used, materials consumed, results, exceptions and disposition.

On this page · 12 sections
- 1What Is a Device History Record?
- 2The Current US QMSR Framework
- 3Historical DHR, DMR and DHF Vocabulary
- 4Define the Record Boundary and Authoritative Index
- 5Record Actual Execution, Revisions and Materials
- 6Keep Results, Exceptions and Human Review Distinct
- 7Labeling, Packaging and UDI Evidence
- 8Worked Example: Quantities, Rework and Label Balance
- 9Electronic Records and Part 11 Applicability
- 10Software Assurance Without Blanket Testing Rules
- 11Retention, Retrieval and Later Reconstruction
- 12Eight Selected Review Questions and Next Steps
How does DHR apply to your shop floor?
Pick your industry and scale — Ask V5 rewrites the definition in your context, gives a worked example, and shows what V5 does on day one.
01What Is a Device History Record?
A Device History Record (DHR) is the organized production history of an identified finished medical device, lot or batch. In operational use, it connects the manufacturing instructions that applied with evidence of the work actually performed, the materials used, the results obtained and the decisions made. It describes execution—not simply what a work order planned or what a design specification intended.
A useful DHR answers a concrete question: what evidence supports the recorded status of these devices? That includes unfavorable evidence. A failed inspection, an unfinished test, a material substitution or a corrected entry belongs in the history with its context; a tidy summary should not make those events disappear.
This reference uses DHR, device master record (DMR) and design history file (DHF) as familiar historical and operational vocabulary, not current QMSR section names. Its regulatory scope is US finished devices intended for human use and the operations a manufacturer actually performs. It does not propose a universal international record format.
02The Current US QMSR Framework
The Quality Management System Regulation became effective on 2 February 2026. FDA explains that the revised Part 820 incorporates ISO 13485:2016 by reference. Current obligations should therefore be assessed under the QMSR, rather than by presenting the former DHR section as current law. See the FDA QMSR overview.
Under §820.1, the regulation covers methods, facilities and controls for design, manufacture, packaging, labeling, storage, installation and servicing of finished devices intended for human use. A manufacturer performing only some covered operations needs to comply with the requirements applicable to those operations. This matters when production history crosses organizational boundaries: a packaging operation and a manufacturer performing assembly through final testing do not necessarily generate the same local evidence.
Section 820.7 identifies the incorporated ISO 13485:2016 edition. Section 820.10(a) requires a documented quality management system complying with applicable requirements of that standard and other applicable requirements of Part 820. Sections 820.35 and 820.45 add specific record and labeling controls. These are regulatory anchors, not a ready-made DHR screen layout. Read the current Part 820 provisions when establishing applicability.
As practical record governance, document which operations your record covers, which evidence comes from another organization and who can retrieve it. Do not fill a template with activities your organization did not perform. Equally, do not let an outsourced operation disappear merely because its detailed evidence is maintained elsewhere. Define the relationship and the evidence needed for your own decisions.
03Historical DHR, DMR and DHF Vocabulary
As of 1 February 2026, former §820.184 required DHRs for each batch, lot or unit to demonstrate manufacture in accordance with the DMR and the then-applicable Part 820 requirements. It allowed the DHR to include information or refer to its location. This is a historical provision, not the current production-record checklist. The dated historical §820.184 is useful for interpreting older procedures and archives.
Its listed information comprised manufacturing dates, quantity manufactured, quantity released for distribution, acceptance records, the primary identification label and labeling used for each production unit, and identifiers and control numbers used. Preserve that historical context when reading an old DHR. Do not assume every item has been restated in identical language in the current framework.
| Familiar Term | Operational Meaning Used Here | What It Does Not Establish |
|---|---|---|
| DHR — Device History Record | Evidence of what happened to identified production subjects. | A planned routing alone does not establish completed work. |
| DMR — Device Master Record | The controlled manufacturing baseline: specifications, instructions and acceptance criteria. | An approved instruction does not establish that a particular run followed it. |
| DHF — Design History File | The history of design decisions and supporting design evidence. | Design evidence does not replace the execution history of a manufactured batch. |
| Work Order | The job being organized, including planned product and quantity. | Planned quantity is not automatically quantity manufactured or released. |
The practical distinction is baseline versus history. A DHR may reference a manufacturing instruction and a design-derived specification without becoming either the master specification collection or the design file. Keep those purposes distinguishable, even if one document-management environment holds all three.
04Define the Record Boundary and Authoritative Index
Treat the DHR boundary as a documented evidence boundary, not necessarily a folder boundary. State which device identities, batches, operations and review checkpoints it covers. A shared test report may support several production groups; an index should identify the applicable portion rather than leaving a reviewer to guess which results belong to which devices.
| Record Relationship | Practical Treatment | Retrieval Question |
|---|---|---|
| Direct production entry | Keep the observation associated with its operation and production subject. | Can the reviewer identify what was done and to which devices? |
| Evidence held elsewhere | Reference a specific controlled record, revision or result set. | Does the reference retrieve the historical evidence, not today's replacement? |
| Derived summary | Identify its source records and calculation or selection basis. | Can the reviewer reconstruct the total or conclusion? |
| Related but separate history | Link later complaint or servicing records without rewriting the original production state. | Can the reviewer distinguish manufacture from subsequent events? |
For each referenced item, a useful index records its identifier, relevant revision or version, scope, authoritative custodian and retrieval location. Where records can change, identify the state supporting the decision being reviewed. A link named latest inspection report is not a reliable substitute for identifying the inspection report actually assessed.
- Identify the SubjectStart with a device or batch identifier and establish the covered population.
- Open the Controlled IndexLocate the operations, evidence sources and review checkpoint relevant to that population.
- Retrieve the Historical BasisRead the executed instruction revisions, material records and original results.
- Follow the ExceptionsConnect failures, pending evidence, corrections, rework and their dispositions.
- Reconstruct the DecisionIdentify what the reviewer assessed, what was authorized and what remained restricted.
A practical retrieval exercise should include a difficult record, not just a clean run. Ask someone who did not assemble the package to follow its references. Missing permissions, an unavailable external attachment or an unexplained summary total are findings for the record owner to resolve—not reasons to substitute memory for evidence.
05Record Actual Execution, Revisions and Materials
The core distinction is between an instruction and an execution. An instruction says what to do; an execution entry says what was done, to what subject, by whom or by which equipment, and when. In a practical record model, connect an operation to the applicable instruction revision and its observations. Avoid recording only that the work order reached a completed status.
Preserve the historical revision used for each operation. A later revision should not silently replace it in a regenerated production record. This does not mean every operation must use the revision current at the instant someone opens the archive, or that every in-progress job must remain on one revision regardless of approved changes. Record an authorized changeover with its affected operations and population, while retaining the earlier basis.
- Material identity: distinguish the required part from the particular component lot or serial actually used.
- Material status: connect use to the status decision relevant at the time, rather than displaying only today's inventory status.
- Quantity: distinguish reservation, issue, consumption, return and scrap. A reservation is not evidence of consumption.
- Execution context: retain equipment identity, observations, units and the applicable method or criterion where relevant.
- Corrections: preserve the original entry, the changed information, the reason and attribution rather than rewriting history.
For example, a reservation for 240 component kits might support a job in which only 232 kits enter assembled devices. The other eight remain unused; they should not appear as consumed simply because they were reserved. If material is removed during rework, record that event separately from the replacement's installation so the final configuration can be reconstructed. These are record-design examples, not universal transaction categories imposed by the DHR name.
06Keep Results, Exceptions and Human Review Distinct
An observed result and an acceptance criterion answer different questions. The result describes what was observed; the criterion describes how it was assessed. In a useful production history, a reviewer can retrieve both, including the method revision, units where relevant and the device or group tested. A pass flag without that context may be insufficient for the review question at hand.
Keep pending, failed and not applicable distinguishable. Pending means the evidence or activity is not yet available or complete. Failed means an assessment found a result unacceptable against the applicable criterion. Not applicable needs a defined basis; it should not become a convenient replacement for missing evidence. A missing result is not a passing result, and a failed result should not be converted to pending merely because rework is planned.
For rework, preserve the originating failure, affected population, authorized action, instruction revision, execution and subsequent assessment. A retest is a new event, not a correction to an accurately recorded earlier failure. Reworked units remain a subset of the original population; counting them again as newly manufactured units distorts the history.
As an accountable review practice, define who assesses evidence, who can authorize disposition and what population each decision covers. Record completion checks can help organize that review, but filled fields do not establish acceptable results. An open issue should remain visible with its impact and status. Whether it prevents a particular disposition depends on applicable requirements and controlled arrangements, not a universal rule that every open record permits—or prohibits—all release.
Do not infer a universal second-person verifier or a signature on every operation from the term DHR. Establish review responsibilities and any required separation of duties through the applicable requirements and documented process. Electronic-signature identification components are a separate subject from the number of human reviewers.
07Labeling, Packaging and UDI Evidence
Current §820.45 requires documented and maintained procedures describing activities for labeling and packaging integrity, inspection, storage and operations under the conditions specified in that section. It also establishes specific examination and documentation requirements. These are current US requirements, unlike the historical §820.184 list. See §820.45 in Part 820.
- Accuracy examination: labeling and packaging must be examined for accuracy prior to release or storage where applicable. The listed items are the correct UDI, UPC or other device identification; expiration date; storage instructions; handling instructions; and additional processing instructions.
- Labeling release: release of labeling for use must be documented in accordance with the incorporated record-control requirements referenced by §820.45(b).
- Mixup prevention: labeling and packaging operations must prevent mixups, including inspection before use to assure correct labeling and packaging as specified in the medical device file. Results of that inspection must be documented.
Separately, §820.35(c) requires the UDI to be recorded for each medical device or batch of medical devices. This is the relevant recording statement here; it should not be converted into a device-class stereotype about lot versus serial tracking. See §820.35.
For practical production history, identify the labeling actually used, its controlled revision, relevant variable information, release evidence and inspection results. Choose supporting evidence that lets reviewers establish identity and the checks performed. Do not turn this into an assumed universal requirement for specimens of every label, machine vision, a particular barcode grade or an independent verifier.
A label inventory balance can provide an additional control when its counting boundary is clear. Define whether counts represent individual labels, sheets or another unit. Separate application, destruction and return, and investigate unexplained differences. The balance below is an illustrative control—not a formula prescribed by §820.45.
08Worked Example: Quantities, Rework and Label Balance
The job plans 240 units. Assembly completes 232 under MI-62 revision D; eight never enter assembly. Consumption records identify 150 kits from component lot C-71 and 82 from C-72. The eight reserved but unused kits are excluded from consumption. The record links the relevant material-status evidence rather than inferring eligibility from a reservation.
At the first functional-check checkpoint under TC-19 revision B, 216 units pass, ten fail and six await testing: 216 + 10 + 6 = 232. Seven of the ten failed units undergo authorized latch-adjustment rework under RW-08 revision A. Six pass the subsequent check; one still fails. The other three originally failed units are scrapped.
| Final Review Population | Units | How the Count Was Derived |
|---|---|---|
| Passed checks; awaiting record disposition | 222 | 216 first-pass units plus six successful rework outcomes. |
| Scrapped | 3 | Three of the ten originally failed units. |
| Failed and held | 1 | One reworked unit still failing the check. |
| Testing pending and held | 6 | The original six pending units; no result has been invented. |
| Total assembled | 232 | 222 + 3 + 1 + 6. The seven reworked units are not added again. |
Label record LB-44 revision E covers 250 labels issued to this run. The exercise assumes the applicable identifier recording, labeling-release evidence and inspection results are linked. Before the production disposition, the label inventory must be reconciled under this fictional procedure.
| Label Category | Count | Counting Boundary |
|---|---|---|
| Issued | 250 | All labels issued to this run, in individual-label units. |
| Applied | 222 | One on each unit that passed the stipulated checks. |
| Destroyed | 9 | Unused spoiled labels, documented as destroyed. |
| Returned | 19 | Unused labels physically accounted for and returned to controlled stock. |
| Balance | 250 = 222 + 9 + 19 | Issued = applied + destroyed + returned. |
The first reconciliation had recorded only 17 returned labels. It therefore accounted for 248 labels and left two unexplained: 250 − (222 + 9 + 17) = 2. A documented recount locates two unused labels inside the return container. The correction retains the original 17, records the corrected 19 and explains the supporting recount. It does not create a fictitious destruction entry to force agreement.
After reviewing the connected evidence and resolved label discrepancy, the designated reviewer authorizes distribution of the identified 222 units under the exercise's stipulated arrangements. The three scrapped units, one failed unit and six pending units keep their separate dispositions. The release is not justified merely because the arithmetic balances. A later MI-62 revision E also does not replace revision D in this historical record.
09Electronic Records and Part 11 Applicability
An electronic implementation is an eDHR, but the software label does not determine Part 11 scope. Section 11.1(b) applies to electronic records created, modified, maintained, archived, retrieved or transmitted under FDA regulatory record requirements, and to specified electronic submissions to FDA. It does not apply to paper records merely transmitted electronically. Begin with the required record and its intended electronic use. See 21 CFR Part 11.
For a closed system, access is controlled by those responsible for the electronic records' content. Applicable §11.10 controls include authorized access, authority checks, accurate and complete copies, record protection and secure, time-stamped audit trails. Changes must not obscure previously recorded information. Section 11.30 addresses open systems with appropriate controls and additional measures as necessary under the circumstances. A software purchasing claim is not a substitute for evaluating these arrangements.
Where Part 11 electronic-signature requirements apply, signed records must show the signer's printed name, signing date and time, and signature meaning; those details belong in human-readable forms as well. Signatures must be linked to their records. Section 11.100 requires unique individual signatures and identity verification before sanctioning them. These are selected controls, not a complete Part 11 checklist.
For non-biometric electronic signatures, §11.200 requires at least two distinct identification components, such as an identification code and password, with rules for their use during signing sessions. This concerns authentication of one signer's identity. It is not a requirement for two people to approve every DHR.
FDA's CSA discussion also distinguishes required electronic validation evidence from routine software logs that are not needed to support a validated state. Not every automatically generated log is therefore automatically a Part 11 record. For electronic terminology, see the eDHR definition; for format choices, use the paper-versus-electronic comparison.
10Software Assurance Without Blanket Testing Rules
FDA's final Computer Software Assurance for Production and Quality Management System Software guidance was issued on 3 February 2026 and supersedes the version issued on 24 September 2025. It provides nonbinding recommendations for production and quality-management-system software, not design-and-development verification or validation of device software functions. See the FDA CSA guidance.
FDA explains that production and QMS software must be validated for its intended use. Its recommended assurance approach assesses software functions and reasonably foreseeable failures. A high process risk exists when failure may cause a quality problem that foreseeably compromises safety; not-high process risk means failure would not cause such a safety-compromising quality problem. The software category alone does not settle that assessment.
Consider the function, safety consequence and additional controls, including meaningful human checks. A display used by a reviewer and a function that determines acceptance without further review may have different consequences if they fail. For an evaluation, document what the reviewer actually checks and whether that check could detect the failure; merely adding a review button does not establish an effective control.
FDA does not assign scripted testing exclusively to high-risk functions or unscripted testing exclusively to other functions. Its guidance describes risk-based selection and combinations of methods. Recommended assurance records cover intended use, risk analysis, activities performed, issues, conclusions, who performed the assessment and when, plus review and approval when appropriate. This is not a universal independent-verifier requirement.
Section 11.10(a) contains a system-validation requirement. FDA's CSA guidance discusses enforcement discretion for specific Part 11 validation requirements, but expressly distinguishes that policy from production and QMS software-validation obligations. Enforcement discretion is not repeal of those underlying obligations. Keep a function's assurance evidence connected to its intended use, rather than treating a generic supplier test package as automatic acceptance of your configured process.
11Retention, Retrieval and Later Reconstruction
Do not derive a universal retention duration from the name DHR, assume that every record is per-unit, or translate device lifetime into an automatic keep-forever rule. Establish the applicable record obligations and document the retention decision in controlled policy. Assess the incorporated record controls and other applicable requirements rather than copying an obsolete retention statement into a new procedure.
For applicable closed electronic systems, §11.10(c) requires protection enabling accurate and ready retrieval throughout the retention period. Section 11.10(e) requires the relevant audit-trail documentation to remain at least as long as the subject electronic records. A retention assessment therefore needs to consider more than the lifespan of a printable summary. See Part 11 record controls.
As practical archive management, identify who maintains the index, historical revisions, attachments, corrections and decision evidence after system replacement or a supplier relationship ends. Test whether the retained package still answers the original review questions. If a summary depends on external data, establish how that data remains interpretable and accessible. Set internal retrieval objectives appropriate to your process; do not present an invented regulator deadline.
For a later complaint or investigation, reconstruct the production state first, then distinguish subsequent servicing, replacements or corrections. Shared material or process exposure can guide further examination, but a shared model name alone does not establish that every device had the same production history. Record unresolved scope questions rather than assuming inclusion or exclusion.
FDA's QMSR FAQ says investigators may review QMS records created before 2 February 2026 and suggests comparative analysis may help demonstrate how earlier records meet QMSR requirements. That supports a careful historical crosswalk, not wholesale rewriting of approved old records. See the FDA QMSR FAQ.
12Eight Selected Review Questions and Next Steps
Use these questions to sample production-record coverage and ownership of linked evidence. They are partial checks, not an overall medical-device readiness assessment, an inspection checklist or a compliance certification. In a buyer evaluation, request demonstrated evidence using test data rather than accepting a feature description.
- Boundary owner: Can the record identify its covered devices, operations and excluded or separately managed activities?
- Document owner: Can a reviewer retrieve the exact instruction and criterion revisions used?
- Material-record owner: Can actual consumption and status evidence be distinguished from reservation or staging?
- Test-record owner: Are original results, units where relevant, failures and pending activities distinguishable?
- Exception owner: Can rework and retest be followed without erasing the original failure or double-counting units?
- Labeling owner: Can label identity, required checks and any inventory discrepancy be explained?
- Disposition owner: Does each decision identify its affected population and supporting evidence, including unresolved restrictions?
- Archive owner: Can the record and its authoritative references still be retrieved after changes or retirement?
For concrete rollout planning, continue to the eDHR implementation guide. Use the eDHR definition for electronic-record terminology and the DHR versus eDHR comparison for paper-versus-electronic intent. This definition establishes what the production history represents; implementation determines how your organization will capture, review and preserve it.
Frequently asked questions
Q.Is DHR still the name of a current QMSR section?+
No. Use DHR as familiar historical or operational vocabulary, not as the title of a current QMSR section. Former §820.184 describes the historical requirement. Since 2 February 2026, current Part 820 uses the incorporated quality-management framework and supplemental FDA provisions. Keeping DHR in an internal procedure can be useful, but map that procedure to applicable current requirements rather than treating its name as evidence of compliance.
Q.How does a DHR differ from a work order, DMR or DHF?+
In the operational distinctions used here, a work order organizes the job, the DMR supplies the manufacturing baseline, and the DHF concerns design history. The DHR records execution for identified production subjects. An approved plan or specification is not evidence that a particular operation happened. The production history should connect the applicable baseline with actual materials, results, exceptions and decisions.
Q.Must there be a separate DHR for every serial number?+
Do not infer a universal per-serial-number requirement from the term. Historical §820.184 referred to each batch, lot or unit; current §820.35(c) requires UDI recording for each medical device or batch. Assess applicable identification and traceability obligations separately. As a practical grouping rule, retain enough identity and evidence relationships to distinguish populations with different results, histories or dispositions.
Q.Must the record contain complete instruction copies and specimens of every label?+
Do not make either an automatic rule based on the DHR name. A practical record can use controlled references to exact historical instructions and supporting evidence. Current §820.45 establishes labeling and packaging controls, labeling-release documentation and inspection-result documentation; this should not be inflated into an assumed universal specimen-of-every-label requirement. Determine the evidence needed for the applicable requirements and your controlled process.
Q.Who reviews the record, and are two signatures always required?+
Assign review and disposition authority through the applicable requirements and documented process. Do not assume a universal independent reviewer or two signatures on every DHR. Where Part 11 applies to non-biometric electronic signatures, its two distinct identification components concern one person's authentication, not two human approvers. In practice, make the decision, its meaning and the population it covers clear.
Q.Can pending tests and failed units be treated as one open-items category?+
They can appear in a shared work queue, but preserve their different meanings. Pending indicates unavailable or incomplete evidence; failed indicates an unacceptable assessed result. Identify the affected units and restrictions separately. In the fictional example, six pending units and one still-failing unit remain held while a separately identified population receives a disposition under the stipulated procedure. The example is not general permission for partial release.
Q.Does an electronic DHR automatically satisfy Part 11?+
No. Assess the underlying required records, their electronic use and applicable system and signature controls. Neither an application label nor a PDF export establishes those controls. A practical evaluation should examine authoritative evidence, access, corrections, signatures where applicable and complete retrieval. The DHR versus eDHR comparison addresses format intent; the eDHR definition addresses electronic terminology.
Q.How long should a DHR be retained?+
Determine the applicable retention obligations and document the result in controlled policy; there is no duration established merely by calling something a DHR. Include the evidence needed to interpret the history, not just its cover sheet. For applicable closed electronic systems, Part 11 requires accurate and ready retrieval throughout retention and retention of relevant audit trails at least as long as the subject records.
Primary sources
- 21 CFR Part 820 — §§820.1, 820.7, 820.10, 820.35 and 820.45; 21 October 2026
- Historical 21 CFR 820.184 — Device History Record; 1 February 2026 only
- 21 CFR Part 11 — Electronic Records; Electronic Signatures; 21 October 2026
- FDA — Quality Management System Regulation Overview
- FDA — Quality Management System Regulation Frequently Asked Questions
- FDA — Computer Software Assurance for Production and Quality Management System Software; 3 February 2026
Further reading
Explore this topic
DHR sits inside 2 overlapping topic clusters in our glossary. Every neighbour is one click away.
Master and executed records that prove a batch or device was made to spec.
Device-specific rules, submissions and the standards that bind them.
Want to see how DHR could fit into your own records and workflows? Explore the related V5 pages or talk to our team about what applies to your operation.
