How to Recover Deleted Files After a Power Outage Shut Down Your PC Mid-Save

The screen went black at 11:42 on a Tuesday morning. I remember the time because the file I’d been editing for three hours (a client’s budget spreadsheet, half-saved) was still open in Excel. The power came back forty seconds later. The file did not.
That’s the scenario this piece is about: not a crash while you were browsing, but a cut that landed mid-write. When the PC dies in the middle of a save, you’re dealing with a very specific kind of damage, and the recovery steps are different from ordinary undelete. Get the order wrong and you can turn a broken file into a gone one.
What actually happens when the power dies mid-save
When you hit save, the application doesn’t just scribble the finished file onto disk. It builds a temp copy, writes it to disk in chunks, then swaps it in place of the original. On NTFS (the standard Windows filesystem) the swap happens through journaling, which is meant to keep the filesystem consistent even if power is lost. The journal protects the directory structure, not your unsaved data inside the file.
So you get one of three outcomes depending on where the cut landed. If the power went before the swap, your original stays intact and the temp file is orphaned. If it went during the swap, the file pointer may still aim at the original but the file is now half-written. If it went after, the file is fine and you just never saw the “saved” confirmation.
Which of those you got is impossible to know from outside. All you can do is stop writing to the disk, then look at what’s actually there.
Step 1: stop using the drive
Every write to the drive marks free blocks as used. If your half-written file’s data is sitting in a block Windows now considers free, the next save fills it with something else and the data is genuinely gone. That’s not a scare tactic. It’s how block reuse works, and it’s the single reason recovery attempts fail.
If the drive is the system drive, shut down and boot from a USB recovery tool. If it’s a separate data drive, unplug it and connect it to another machine as a slave. If you don’t have a second machine, at minimum stop opening applications and don’t install anything new until you’ve copied the suspect files somewhere else.
Where the half-saved version lives
Before you reach for a recovery utility, look for the application’s temp and autosave files. Most desktop software drops them in your AppData or user folder, and they’re often still readable. Microsoft Office keeps recent versions in its AutoRecover cache; LibreOffice has a clearly labelled backup folder. Search for the filename plus a tilde, or sort your temp folder by modification time to catch the file with the right timestamp.
I’ve recovered a half-finished InDesign file twice this way, no utility needed. The temp file was ugly but complete enough to open and resave. Worth trying before a scan, because a scan is slower and has a smaller success window.
Step 2: pick a recovery tool that fits the file type
The tool you choose depends on what you lost. This is where most guides get vague, so here’s how to think about it:
- For deleted or overwritten files on NTFS drives (non-SSD), a standard undelete tool will do. Recovery Toolbox’s File Undelete is a free example built for NTFS, though it doesn’t work on SSDs at all, which matters because most modern laptops have one.
- For a broken file format (a corrupted Excel workbook, a damaged RAR archive, a half-written Outlook data file), a format-specific repair tool handles it better than a generic undelete. Recovery Toolbox produces dedicated repair tools for each of those formats, and they’ll often salvage the contents of a damaged file the general undelete utilities can’t see.
- For anything on an SSD, expect a real limitation. SSD controllers run TRIM, which wipes blocks the moment they’re marked free, so undelete has nothing to find. Recovery Toolbox’s own documentation says File Undelete doesn’t support SSD storage. The tools that do work with SSDs often only work by catching data before TRIM runs, which you don’t control.
If you can’t tell whether your drive is spinning or solid-state, check the model number. Most laptops made after around 2018 are solid-state. If yours is, and the file was saved, then lost in a crash, your odds are lower than you’d like. I’d still try: TRIM doesn’t always fire immediately, and it doesn’t fire at all if the drive was removed from the machine before Windows could run it. But I wouldn’t spend good money on a scan without checking that first.
Step 3: scanning and the one rule that matters
Whatever tool you choose, scan the drive in read-only mode and recover to a different physical disk. Never recover a file back to the drive you just scanned from. That’s the fastest way to overwrite the same file you’re trying to save.
If your tool offers a preview mode (most do), use it to confirm the file you want is actually among the recoverable items before you pay for the full version. Half the utilities out there advertise recovery and then charge you to see anything. Check the preview.
Expect the recovered version to be broken. If the crash happened during the save, the recovered data will be truncated or corrupted in some way because the write wasn’t completed. That’s not a failure of the tool. Recovery tools rebuild what’s on the disk; they can’t invent the four paragraphs you never typed.
Step 4: after the fact, recover the unsaved work
If the file that got hit is one you can rebuild from history (a document with a cloud version, a source file committed to a repository, a spreadsheet with exported copies), the recovery scan is a fallback, not the first move. Check the cloud version history or the version control log before you touch the broken drive. Those recoveries are free and complete, and they don’t care that the power died.
If you have a backup (and I mean a real backup, ideally a disk image that lets you restore the whole machine to a moment before the crash), restoring the previous copy beats any recovery tool on every metric. The industry’s own reviewers, including PCWorld’s breakdown of Windows backup software, are blunt about the built-in Windows utilities: there’s a patchwork of File History and legacy imaging tools there, and none of them are suitable for quick disaster recovery. Third-party imaging software is where restore-from-image actually works reliably. If you don’t have an image, set one up today, because you now know what a forty-second power cut costs.
What I’d do in your chair right now
First, stop writing to the disk. Second, hunt through the temp and autosave folders of the crashed application, because there’s often a complete copy sitting there. Third, check your cloud or version control history. Fourth, if all three are empty, run a read-only recovery scan, preview the results before paying, and recover to a different drive. If your drive is solid-state, run the scan anyway, but don’t expect a miracle.
The version of this that bothers me is the one where someone keeps working on the same machine for two more hours before trying recovery. By the time they open a tool, the blocks are full of someone else’s data. That’s the loss that hurts.
The half-saved file from that Tuesday was gone. The original, though, was still sitting in the folder I’d been editing. I’d forgotten to overwrite it with the newer version. It took me an afternoon to retype the changes, and now I save every ten minutes (Ctrl+S without thinking, a habit I’ve kept for years). Cheap insurance. The real fix, an image on an external drive, took another week to set up, but that one genuinely stops the next crash from being interesting.