What it is and why it matters

When a file is deleted the operating system usually removes the pointer to it, not the content. The clusters are marked free and the data sits there until something else is written over it. Recovery is the craft of reading those orphaned clusters back into files, using whatever the file system left behind: an unused directory entry, a journal record, a copy in a snapshot or, failing all of those, the recognisable header of the file itself.

Data recovery and evidence recovery use the same techniques and reach different standards. A recovery shop works on the customer's drive, tries whatever gets results and hands back the files. A forensic examiner works only on a verified image, records every method, keeps the recovered items linked to the sectors they came from and reports how confident the recovery is. A carved photo with no file name and no timestamp is still useful, but it must be described honestly.

Recovery also has an end. Data that has been overwritten is gone, and so is data on a trimmed SSD or behind encryption without the key. NIST SP 800-88 Rev. 2 makes the point from the other side: a single proper overwrite defeats recovery on a hard drive. An examiner who understands sanitisation knows when to stop searching and say so.

Types of digital evidence

  • Deleted Files With Intact Records

    Files whose NTFS $MFT entry or FAT directory entry still exists but is marked unused. Name, size, timestamps and cluster pointers are all available, so recovery is exact.

  • Journal Copies

    Metadata and sometimes data preserved in the NTFS $LogFile and USN journal, the ext4 journal or the APFS checkpoint history, showing a file's name and changes after the main record is gone.

  • Carved Files

    Files rebuilt from unallocated space by recognising headers and footers such as JFIF for JPEG or %PDF for PDF. No name or timestamp travels with them.

  • Snapshots and Shadow Copies

    Earlier versions of files kept by Volume Shadow Copy, APFS snapshots, Time Machine or NAS snapshots that still hold what the live volume has lost.

  • Recycle Bin and Trash Records

    Windows $I and $R files and macOS Trash metadata that record the original path and deletion time of items moved there.

  • Slack and Fragment Data

    Partial content in the slack of allocated files and in isolated unallocated clusters, enough for keyword hits and sometimes for partial documents.

  • Recovered Media From Failed Drives

    Data read from a mechanically or electronically failed drive after head swap, firmware repair or controller-level recovery.

  • Application Caches and Thumbnails

    Thumbnail databases, browser caches and sync client databases that keep small copies of files deleted elsewhere.

Devices and data sources

  • Hard drives from computers, DVRs and external enclosures
  • SSDs and NVMe drives, with the limits that TRIM imposes
  • USB pen drives and memory cards, including those with failed controllers
  • RAID and NAS volumes after virtual reconstruction
  • Mobile phone file system extractions and chip-off dumps
  • Virtual machine disk files and their snapshots
  • Cloud sync client databases and local caches
  • Backup media and archives that hold earlier versions

How an examination runs

The order matters. Each step protects the one after it.

  1. 01
    Work From the Image

    Every recovery runs against a verified forensic image or a hardware-level clone, never against the original. If the drive is failing, the image is made with a recovery-aware tool first.

  2. 02
    Read the File System Records

    Parse the $MFT, FAT tables, ext4 inodes or APFS trees for entries marked deleted but still intact. Recover these first, since they carry names and timestamps.

  3. 03
    Use the Journals

    Read the NTFS $LogFile and USN journal, the ext4 journal and APFS checkpoints for records of files that have lost their main entry, to recover names, paths and times of deletion.

  4. 04
    Check Snapshots and Bins

    Mount Volume Shadow Copies, APFS and NAS snapshots and read Recycle Bin and Trash metadata before carving, because they often give complete files with context.

  5. 05
    Carve Unallocated Space

    Run signature-based carving for the file types that matter to the case. Validate carved results by opening them, and expect fragmented and partial files.

  6. 06
    Record Provenance

    For every recovered item record where it came from: the file system entry, the journal record or the sector range carved. Mark carved items as having no original name or timestamp.

  7. 07
    Grade Confidence

    Separate exact recoveries from reconstructed and partial ones in the report, and say what could not be recovered and why.

Tools used in this discipline

Independent profiles: what each tool does, where it stops and what you need to know before using it.

Terminology

Logical Recovery
Recovering data from a readable device by reading file system structures and unallocated space. No hardware repair is involved.
Physical Recovery
Making a failed device readable again, by head swap, firmware repair, controller-level access or chip-off, so that an image can be taken.
File Carving
Reconstructing files from raw data by header and footer signatures, without help from the file system.
Fragmentation
A file stored in non-contiguous clusters. Carving assumes contiguity, so fragmented files come back corrupted or incomplete unless the file system records survive.
Overwrite
New data written to a sector that held old data. The old content is not recoverable by software, and on modern drives not by any practical means.
Journal
A log the file system keeps of metadata changes, used to recover after a crash. It also preserves records of deleted files for a while.
Snapshot
A point-in-time view of a volume kept by the file system or backup software. Deleted files often survive in a snapshot taken before the deletion.
Provenance
The documented origin of a recovered item: which record, journal entry or sector range it came from.
Secure Erase and Cryptographic Erase
Drive commands that erase all cells or destroy the encryption key. After either, there is nothing to recover, and recognising this saves wasted effort.

Challenges and limits

What makes it hard

  • Overwritten data is unrecoverable, and the only question is how much was overwritten
  • TRIM on SSDs removes deleted content on the drive's own schedule, so recovery from flash is unpredictable
  • Encryption without the key leaves nothing to recover even from a perfect image
  • Fragmented files defeat carving and come back as mixed or truncated content
  • Carved files have no name, path or timestamp, which limits what can be said about them
  • Failing drives may allow only one imaging attempt, so the first method must be the right one
  • Recovery tools produce false positives, and every important recovered file must be opened and checked

Keeping the evidence intact

  • Recover from a verified image or clone, never from the original device
  • Record the source of every recovered item at the sector or record level
  • Hash recovered files individually and keep them with the record of how they were recovered
  • Mark carved and partial recoveries clearly and never present them with invented names or dates
  • Document every tool, version and setting so the recovery can be repeated

Standards and legal considerations

  • ISO/IEC 27042:2015

    Guidance on the analysis and interpretation of digital evidence, including how to present reconstructed data and state confidence.

    Read the guide
  • ISO/IEC 27037:2012

    The preservation rules that keep the original intact while recovery runs on a copy.

    Read the guide
  • NIST SP 800-88 Rev. 2

    Media sanitisation guidance. It defines what makes data unrecoverable, which tells an examiner where recovery honestly ends.

    Read the guide
  • SWGDE Best Practices

    SWGDE guidance on examination and reporting that covers recovered and carved data.

    Read the guide
  • Digital Evidence Lifecycle

    Recovery sits in the examination stage, between acquisition and analysis, and its output must carry the same custody as the original.

    Read the guide

In practice

  • Deleted Invoices With Names Intact

    An employee deleted a folder of invoices the day before an audit. The $MFT still holds the entries marked unused, with names, sizes and cluster pointers. The examiner recovers all of them exactly and shows from the USN journal the time each was deleted.

  • Photos From a Formatted Card

    A memory card was quick-formatted. The FAT tables are blank, so the examiner carves the data area for JPEG headers and recovers several hundred photos. The report explains that the file names are lost but that EXIF timestamps inside each photo survive.

  • Nothing to Recover

    A drive handed in as evidence turns out to have been sanitised with the drive's secure erase command, confirmed by the SMART data and a uniform pattern across every sector. The examiner reports that no data is recoverable and why, rather than running tool after tool.

This page is educational material, not legal advice. Laws and standards change, commence on different dates and are interpreted differently across jurisdictions. Always read the official text linked on this page and take advice from a qualified professional for a real matter.

Learn it at DFRTI

Free pages explain. These courses teach it on real devices with a mentor.

Questions people ask

What is the difference between recovering a file and carving one?

A recovered file comes back through its file system record, with name, path and timestamps. A carved file is rebuilt from raw data by its signature and arrives with none of that. Both are useful, but they are reported differently.

Can overwritten data be recovered?

No. Once a sector has been rewritten the old content is gone. Claims of recovering overwritten data from modern drives are not supported by evidence, and NIST SP 800-88 Rev. 2 treats a single overwrite as sufficient for clearing a hard drive.

Why does recovery on an SSD give so little?

Because the drive erases or zeros trimmed blocks on its own. Unless TRIM was off or the drive had no power since deletion, the deleted content is usually gone. Journals, snapshots and caches become the main source.

Is a data recovery company's output acceptable as evidence?

It can be, if the drive was handled under custody, the work was documented and the recovered items can be tied to their origin. Without those controls the output is useful information but harder to defend.

More in Devices and storage

Go deeper with DFRTI

Free pages explain the ground. Structured courses, lab work and a mentor take you from knowing about it to doing it on a case.