MakeUseOf's Gavin Phillips recently covered the problem. He argues that fake cards turn up on Amazon, eBay and most other big shopping sites, and that testing a card before you rely on it is the only way to be sure. The technique is sound. Several points need sharpening, though, especially around speed, data loss and which tools can destroy data.
Why fake cards fool Windows
TP-Link's support documentation describes the most common fraud as the reduced-capacity card. A 32GB card, for example, can show up as 128GB on your computer. Because the capacity is falsified, space beyond the real capacity can't be used normally, and files stored past that point become unavailable.
The failure is quiet. The RMPrepUSB tutorial, written by the team behind FakeFlashTest, explains why. Spotting a tampered controller takes a tool that doesn't defer to the operating system to work out the storage size. The RMPrepUSB page describes two behaviours:
- Type A: the card accepts all your files and lists every filename. Anything past the real capacity is corrupt, probably all zeros or all ones.
- Type B: the card works until you exceed its real size. The next write then lands at the start of the card, where the file directory lives. That corrupts the directory and makes every file inaccessible.
The RMPrepUSB author adds that most fake drives ship formatted as FAT32. That filesystem fills the card from the bottom up, so the fraud stays hidden until the real memory is full.
The pre-purchase checks: clues, not proof
Phillips gives a few warning signs. None of them proves anything on its own.
- Brand mismatch. His example is a "Xiaomi" 512GB microSD card. He says Xiaomi doesn't make microSD cards, so a buyer would likely get an unbranded card with far less capacity than advertised.
- Generic brands. Unfamiliar names that promise huge capacity at tiny prices, often mimicking SanDisk or Samsung.
- Model numbers that don't match. Phillips says a "Verbatim 44082" listing he found on Wish seemed to be a higher class of card than the real product. He says the matching card is a Verbatim 44010. He didn't buy one, so he can't say what arrives.
I could partly check the model-number approach. Verbatim's own product page lists the 44082 as a 16GB Premium microSDHC card with a UHS-I interface, U1 rating, V10 video speed class and read speeds up to 80MB/s. I couldn't independently verify Phillips's account of the Wish listing or what a buyer would receive. Treat his examples as his reporting.
The practical lesson holds regardless. Compare the exact model and capacity against the manufacturer's specifications. Buy from an authorized seller. TP-Link's guidance says the same: buy from an authorized seller, be wary of unusually cheap cards, inspect packaging, and avoid used or refurbished cards. Phillips prefers Amazon to Wish or Temu, but he admits Amazon is far from perfect. A marketplace name guarantees nothing.
Testing a card on Windows
A write-and-verify test is the one check that doesn't take the card's word for anything. It writes a known pattern across the whole card, then reads every byte back and compares. A card that reports more space than it has will fail where the real flash runs out.
H2testw, the thorough option
H2testw is the long-standing Windows standard. The steps below follow the guidance from TP-Link and the 3DS hacks guide, which agree with each other:
- Back up anything on the card you want to keep.
- Insert the card and run h2testw.exe. Choose English if prompted.
- Click Select target and pick the card's drive letter. Check this twice. Selecting the wrong drive can wipe it.
- Make sure all available space is selected.
- Click Write + Verify and wait.
What success looks like: the test shows "Test finished without errors", and you can delete the .h2w files from the card. A fake card instead reports errors and shows how much data it actually wrote and read back.
Time and caution:
- Expect hours on a large card, since the test writes and then rereads the entire capacity.
- Don't interrupt it.
- Test a new card, not one holding files you care about. The RMPrepUSB author warns that filling a fake card can corrupt files already on it.
- A good USB 3.0 card reader helps. The RMPrepUSB page notes that cheap USB 2.0 readers and integrated laptop readers can be notoriously slow.
FakeFlashTest, the fast option
FakeFlashTest comes from the RMPrepUSB developers. Its developer is clear about its limits. A program that scans only part of the card gives an incomplete picture. That makes FakeFlashTest a quick screen, not a verdict on the whole card. The developer says it does not test every bit of every byte. It looks for missing chunks of memory, which makes it much faster. On a fake 16GB USB drive it reached a diagnosis in 27 seconds.
Phillips describes two modes:
- Quick Size Test: writes and reads 512 bytes at random places across the drive and logs any failure.
- Test Empty Space: writes and reads all available space, then compares the results.
RMPrepUSB also describes a separate RMPrepUSB Quick Size Test that destroys all data on the device and needs a reformat afterward. That warning applies to that specific test. The RMPrepUSB page also says that if a quick test passes, you should reformat the card and run H2testw to check the memory more thoroughly.
One more quirk from the developer: if the Quick Size Test fails every device you try, security software or a driver may be interfering. The developer suggests running it in Safe Mode.
Mac and Linux: F3
Windows isn't the only platform with a fix. F3, short for Fight Flash Fraud, works on Linux and macOS. The standard workflow uses f3write to fill the card with test files and f3read to check them. It's a command-line tool, which Phillips notes isn't to everyone's taste.
- macOS GUI: F3XSwift wraps f3write and f3read. Its README lists macOS 13 (Ventura) or later, including macOS 26, and says it runs natively on Apple Silicon and Intel Macs. The steps are to select the card, press Test, grant access, and wait for the write and then read phases. The README warns this can take several hours for larger or slower cards. Phillips mentions F3X, an older macOS GUI, but F3XSwift is the one I could verify as currently maintained.
- Linux GUI: Phillips points to F3-qt. I didn't verify its current maintenance status.
F3 also includes f3probe, a faster test that runs against an unmounted device. It is destructive. The project documents that it erases existing data, so it isn't a casual shortcut.
What a failed test means
- Don't use the card for anything important. A failure can mean counterfeit capacity or simply defective media. The test can't tell you which.
- Don't expect recovery. Louis Rossmann's repair shop writes that data written past the real capacity was discarded or overwritten by the controller, and no lab can recover it.
- Shrinking it isn't a fix. RMPrepUSB notes you can reformat a fake card to its smaller real size. It adds that the memory is often faulty or likely to fail later, so keep it away from anything that matters. Phillips says fake cards can still be reused for low-stakes jobs.
- Report the seller and ask for a refund. The RMPrepUSB author recommends checking a seller's feedback for complaints about corrupted data before you buy.
The short version
- A reported capacity proves nothing. Only a full write-and-verify does.
- On Windows, H2testw is the thorough test and FakeFlashTest is the quick screen.
- Back up first, test new and empty cards, and block out a few hours for big ones.
- Cheap price, odd brand and mismatched model numbers are reasons to be suspicious. They aren't proof.
- A card that fails shouldn't hold anything you can't afford to lose.
This is the kind of chore that feels pointless right up to the day it saves a holiday's worth of photos. If a card passes in H2testw, you can trust its capacity. Speed and long-term durability are separate questions.