Your Dash Cam Says It Is Recording. The Card Died Months Ago: Why Dash Cam Memory Cards Fail in India, and How to Check Yours This Week (2026)

The dash cam is powered on. The screen shows the road. The little red dot is blinking. Everything looks exactly the way it looked the day it was fitted.

Then a bike cuts across you on Hosur Road, there is a scrape and an argument, and you pull the card to get the clip. The last usable file is from March. The camera has been going through the motions for five months and writing nothing that survives.

This is the most common dash cam failure in India, and it is not a camera fault. It is the memory card, which is the one component nobody thinks about, almost nobody checks, and which is usually the cheapest part of the entire setup. This article covers how to check yours this week, why cards die, what the endurance numbers on the box actually mean for a 4K or three channel camera, and how to tell whether the card you bought is even real.

Check your card this week, before you need it

Do this now. It takes ten minutes and it is the whole point of the article.

1. Pull the card and put it in a computer. Look at the newest file's date. If the most recent clip is not from your last drive, the camera has stopped writing and the display has been lying to you.
2. Play three or four clips, including the oldest one on the card. Look for freezing, blocky squares, files that will not open, or clips that end early. Corruption usually shows at the tail of a file first.
3. Check that locked event files actually exist. Brake hard once in a safe spot to trigger the G sensor, then confirm a protected file appeared. A card that records normally but cannot write a locked file is failing in the worst possible way, because event files are the only ones that matter after a collision.

If any of those three checks fail, replace the card. Do not try to nurse it through another month. A card in this state has already cost you nothing so far only because you have been lucky.

Why loop recording destroys ordinary memory cards

A dash cam uses storage in a way almost nothing else does.

Your phone writes a photo and leaves it there for two years. A dash cam fills the card completely, then goes back to the beginning and overwrites the oldest file, then does it again, continuously, for as long as the engine runs. Flash memory cells tolerate a finite number of erase and rewrite cycles, and loop recording is a machine for consuming those cycles as fast as physically possible.

Endurance cards exist for exactly this workload. They use flash and controllers chosen for sustained rewriting rather than for peak burst speed, and they carry a published endurance rating, which ordinary cards do not. The card bundled free with a camera, or the Rs.400 card off a marketplace listing, is usually a standard consumer card wearing a sticker.

That is the standard explanation, and it is correct as far as it goes. It is also where most articles stop, which leaves the interesting question unanswered: how long does an endurance card actually last in your camera?

What 43,800 hours actually means, and why it is not your number

Samsung rates the 128GB PRO Endurance card at 43,800 hours. SanDisk rates the 128GB MAX ENDURANCE at 60,000 hours and the 256GB at 120,000 hours. Those figures get quoted constantly and almost always without the condition attached to them.

Samsung's own announcement states the test condition plainly: Full HD 1920 x 1080 video at 26 Mbps. That is the fine print that changes everything, because the rating is really a statement about total data written, not about hours.

The arithmetic is straightforward. At 26 Mbps a camera writes about 11.7 GB every hour. Multiply by 43,800 hours and the card's rated life works out to roughly 512 TB of writes, which is about 4,000 full rewrites of a 128 GB card. That number is the real specification. The hours are just one way of expressing it, and they only hold if your camera writes at the same rate as the test.

Most cameras Indian buyers are shopping for do not. Here is the same 512 TB budget divided by realistic write rates for different camera types.

To be clear about what that table is: it is arithmetic derived from Samsung's published rating and its stated test condition, using typical bitrates for each camera class. It is not a bench test, and real bitrates vary by model and by how much motion is in the frame.

The conclusion still holds with a wide margin for error, and it is the opposite of what most buying guides imply. For a car that gets driven a few hours a day, a genuine endurance card will not die of write exhaustion. It will outlast the camera.

Which raises the obvious question. If the cards are rated for decades, why do so many of them fail inside a year?

Parking mode is the setting that changes the arithmetic

Everything above assumed the card only works while you are driving. Parking mode removes that assumption.

A camera wired for 24 hour monitoring can be recording 8,760 hours a year instead of roughly 730. That is a twelvefold increase in writes, and it collapses the numbers from decades into something you will actually live to see.

That is the honest case for replacing cards on a schedule, and it applies specifically to people who hardwire their cameras. If you have fitted something like a hardwire kit for 24 hour parking monitoring, budget for a new card every two to three years and treat it as a running cost of the feature, in the same way you would treat wiper blades.

One important qualifier. The figures above assume continuous recording while parked, which is the worst case. Many cameras use motion triggered or time lapse parking modes that write far less, and those sit somewhere between the two columns. Check which mode yours is actually in, because people frequently assume they are on motion detection when the camera shipped set to continuous. The parking mode guide covers how the modes differ.

Heat is the other half of the Indian problem

A car parked in the open in Bengaluru or Chennai in May turns its windscreen area into an oven, and the card sits in the hottest part of the cabin, stuck to the glass, often in direct sun.

Samsung specifies an operating range of -25C to 85C for the PRO Endurance line. That is a genuinely wide range and it covers Indian conditions, which is precisely the point: the branded endurance cards publish a temperature specification and are built to it. Cheap unbranded cards mostly do not publish one at all, and a card with no stated thermal limit in a dashboard that reaches 75C is an unmanaged risk.

Heat does not usually announce itself. It shows up as a card that was fine all winter and starts dropping files in April, or a camera that reboots on long afternoon drives. If your failures are seasonal, heat is the first thing to suspect.

The counterfeit card problem, and the test that catches it

This is where a lot of Indian failures actually originate, and it connects to a pattern we have written about before with counterfeit dash cams and fake audio brands.

A counterfeit memory card is not simply a lower quality card. It is a small card reprogrammed to report a large capacity to whatever device reads it. A card with 8 GB of real flash can announce itself as 128 GB. Your computer shows 128 GB. Your camera shows 128 GB. The file browser looks completely normal.

The failure only appears once writing passes the real capacity. At that point the card silently overwrites or discards data while continuing to report success, which produces exactly the symptom this article opened with: a camera that looks like it is recording and a card with nothing usable on it. Reformatting does not fix it, because the fake capacity is programmed into the controller rather than stored in the filesystem.

Fake cards sold as Samsung and SanDisk have been reported by buyers on Indian marketplaces, so buying a familiar brand name is not by itself protection.

The test: run a full write and verify pass using a tool such as h2testw or the F3 utilities. These fill the card to its claimed capacity and then read every byte back to confirm it survived. A genuine card passes. A fake fails at the point where the real flash runs out. The test takes a few hours for a large card and it is the only method that actually proves capacity. Browsing the card, formatting it, or checking properties proves nothing.

Do this the day the card arrives, while a return is still straightforward.

What to actually buy, and what to ignore

Four things matter, and the marketing on the front of the pack is mostly not among them.

• Buy a card sold specifically as high endurance. The words to look for are endurance, dash cam, or surveillance. A fast card intended for cameras and drones is optimised for burst speed, not for being rewritten forever.
• Match the speed class to the camera, not to the highest number available. Class 10 with U1 is the floor for most cameras. For 4K or multi channel, look for U3 or V30. These classes describe a guaranteed sustained write floor, which is the number that matters for continuous recording, unlike the large read speed printed on the front.
• Check your camera's supported capacity before buying. 70mai, for example, documents support for cards from 32GB to 256GB with Class 10 or U1 as the minimum. Putting a larger card in a camera that does not support it produces failures that look exactly like a defective card.
• Size the card to your usage, not to the maximum. Bigger cards hold more history and spread writes across more cells, which helps. But a 4K camera fills any card quickly, so if the goal is keeping several days of footage rather than several hours, go up a size.

Buy from a seller who will take a return, and run the verification test immediately. We deliberately are not quoting rupee prices here because card pricing moves constantly and a stale number is worse than none.

The maintenance schedule that prevents most failures

Three habits, and they cover almost everything.

1. Format in the camera, every one to three months. Monthly if you drive daily, quarterly if the car sits. Use the camera's own format function rather than formatting on a computer. The camera writes a filesystem structure suited to loop recording, while a computer applies a general purpose one that the camera then has to work around.
2. Copy off anything you want to keep, before you format. Locked event files are still on a card that gets wiped. If a clip matters for a claim or a complaint, get it onto a phone or laptop the same day.
3. Replace on a schedule if you run parking mode. Every two to three years for a hardwired 4K or multi channel setup. For a camera that only runs while driving, replace when it shows symptoms rather than on a calendar.

Also worth doing once: confirm the camera is actually mounted and configured so that the footage is usable in the first place. A perfectly healthy card recording a badly aimed camera still leaves you with nothing, which we covered in the mounting and settings guide and in the breakdown of why number plates blur at night.

We sell the cameras. We do not sell the cards.

Worth stating plainly, because it explains the tone of this article. Our catalogue runs to more than thirty dash cameras, from a Rs.1,990 single channel unit up to a Rs.31,999 three channel 4K system. We stock no memory cards at all.

So there is nothing to sell you at the end of this. Buy the card wherever you like. The reason for writing it anyway is that a failing card presents itself as a failing camera. A camera gets blamed, returned, or replaced when a Rs.400 card was the actual fault, and that helps nobody. Checking the card first takes ten minutes and settles the question.

If your camera itself is the problem rather than the card, that is a different conversation and one worth having properly. The footage is only worth what it can prove, which is the whole reason a dash cam earns its place in an insurance claim.

Quick questions

Why did my dash cam SD card fail?

Almost always one of four causes: a counterfeit card that never held its claimed capacity, an ordinary consumer card used in a device that rewrites storage continuously, sustained dashboard heat, or 24 hour parking mode multiplying the write load twelvefold. Genuine endurance cards very rarely fail from write exhaustion in normal daily driving.

How long does a dash cam SD card last in India?

For a car driven a few hours a day, a genuine endurance card should outlast the camera. With continuous 24 hour parking mode on a 4K camera the realistic figure is under four years, and around two years for a three channel setup. Heat and counterfeit cards shorten all of these considerably.

Do I really need a high endurance memory card for a dash cam?

Yes, though not for the reason usually given. The value is that endurance cards are designed for sustained rewriting and publish a rated temperature range, which matters in Indian conditions. It is less about a normal driver exhausting write cycles, which the arithmetic says is unlikely, and more about buying a card built and specified for the job.

How do I check whether my memory card is fake?

Run a full write and verify test with a tool such as h2testw or F3. A counterfeit card reports its claimed capacity to every device and only corrupts data once writing passes its real size, so formatting, browsing files, or checking properties will not reveal it. Only writing to full capacity and reading it back proves the card is genuine.

How often should I format my dash cam SD card?

Every one to three months, using the camera's own format function rather than a computer. Monthly suits daily drivers and quarterly is enough for a car that sits. Formatting in the camera writes a structure suited to loop recording. Copy off any footage you need to keep first, because formatting erases locked event files too.

What size and speed memory card should I buy for a 4K dash cam?

Look for a high endurance card rated U3 or V30 for 4K or multi channel cameras, and check the maximum capacity your camera supports before buying. 70mai documents support for 32GB to 256GB with Class 10 or U1 as the minimum. The sustained write class matters more than the large read speed printed on the front of the pack.