The endurance rating on your memory card is capacity times a constant

There is no endurance badge

Every speed mark on a memory card is a defined thing. The SD Association writes the C, U, V, A and E classes, each one is a floor the card has to hold, and anyone with a bench can check it. I have written up what each of those marks certifies, and the short version is that the printed ones are honest and usually conservative.

Endurance is the exception. There is no endurance mark, no endurance class, no standard test and no certification. "High Endurance" is the name of a product line, the way "Extreme" and "Ultra" are names of product lines. What fills the space where a standard would be is a number in hours, and that number is the subject of this article.

SanDisk says as much in its own words. A white paper it published in January 2026, written for the camera makers who buy its industrial cards, explains that consumer cards "typically do not include specified write endurance ratings," and that the commercial video lines express endurance "in thousands of hours of video recording at a specific bit rate, rather than in total bytes written (TBW)," because that "makes the endurance rating easier to understand for non-technical users." That is the vendor describing the simplification I am about to take apart, and it is a fairer description than I could have written.

Two things need saying before the arithmetic, and both stay true to the end. First, a linear ladder proves that the rating carries no information beyond the capacity. It does not prove the cards are bad, and the ones I have measured are not. Second, nothing here is a lifespan measurement. A test that would mean something takes months on a dedicated rig, I have not run one, and the honest thing a bench can offer is at the bottom of the page.

Three ladders, one multiplication

Here are the published ratings, straight from the datasheets and the newsroom page, with one column added: the rating divided by the capacity.

SanDisk High Endurance

SanDisk MAX Endurance

Samsung PRO Endurance

Not one rung on any of the three ladders deviates by an hour. Samsung's is the most transparent about how it was built: 17,520 hours is two years of 8,760, and each doubling of capacity is a doubling of years, so the 128GB card is rated for exactly eight years of recording around the clock and the 256GB for sixteen.

That is the whole finding, and it needs no lab. The rating is not a measurement of any capacity. It is the capacity multiplied by a constant the line chose, so when you compare a 128GB and a 256GB card within one line the hours tell you exactly what the gigabytes told you, and when you compare two lines the hours tell you the ratio of two constants. A 64GB SanDisk High Endurance is rated for 5,000 hours because it is 64 gigabytes, not because a 64GB card was run for 5,000 hours.

Transcend, for what it is worth, rates its High Endurance 350V line in terabytes written rather than hours, 85 TB at 64GB and 170 TB at 128GB according to its product literature, which I could not verify against a datasheet because the company's document host would not complete a secure connection on either day I tried. It is the one line in this article whose unit needs no bitrate to mean something, and it is linear in capacity too.

A 64GB card is rated for 5,000 hours because it is 64 gigabytes. Nobody ran a 64GB card for 5,000 hours to find that out.

The number that gives the hours meaning is on the box, not in the datasheet

An hours rating is a bytes rating in disguise. To turn it back into bytes you need the bitrate it assumes, and here is where each source stands on stating one.

I text-extracted both SanDisk datasheets and searched them. Neither contains the string "Mbps" or "Mb/s" anywhere. The condition is on the retail packaging, in the fine print on the back of the box, transcribed here from the box of the exact 128GB card this bench has measured:

The MAX Endurance 128GB box in the corpus prints the same sentence with "Actual hours of video saved less." in front of it, so both SanDisk lines run on the same 26 Mbps assumption, and Samsung publishes the identical figure. That makes the three lines directly comparable once you know the number. You just cannot learn it from the document an engineer would cite. It lives on the part of the product that goes in the bin.

It also means a card without a box has no condition at all. The SanDisk High Endurance 512GB I benched came out of a dash cam after months of service, with no packaging, so its 40,000 hours arrive from the datasheet with a resolution and nothing else. Its review states the rating as a trusted claim, not an endorsed one, and that is the only honest way to carry it.

The headline belongs to the biggest card in the line

The SanDisk High Endurance datasheet opens with "Designed to record for up to 40,000 hours," and the qualifier that makes it true, for 512GB, is inside footnote 1:

40,000 hours is the 512GB figure. The 32GB card in the same line is rated for 2,500. The headline is sixteen times the smallest capacity's own rating, and because it is the line's headline, it is the number that gets attached to listings for every capacity in the line. A shopper looking at a 32GB High Endurance card under a banner that says 40,000 hours is looking at a figure 16x that card's own datasheet row.

Credit where it is due: the MAX Endurance datasheet handles this better. Its headline bullet reads "up to 120,000 hours, or over 13 years (for 256GB only)," with the qualifier in the sentence rather than in a footnote. Same ladder shape, more honest sentence.

Turn the hours into terabytes and the constant into cycles

Once you have the bitrate, the hours become bytes. At 26 Mbps a card takes 3.25 megabytes every second, so multiply the rated hours by 3,600 seconds and by 3.25, and you have the number of megabytes the rating is promising the card will absorb. Divide by the capacity and you have how many times the whole card gets written over.

Now those rewrite counts can be read against something SanDisk itself published. The same January 2026 white paper lists three of its industrial microSD lines with their cell type and their rated program/erase cycles: a 15 nm planar MLC part "up to 3K P/E cycles capable," a BiCS3 64-layer 3D TLC part, also "up to 3K P/E cycles capable," and a BiCS5 112-layer 3D QLC part "up to 1K P/E cycles capable." Those are the ceilings SanDisk puts on its own industrial silicon, by cell mode.

Read the two tables together. SanDisk High Endurance at 914 rewrites sits comfortably inside a 3,000-cycle TLC budget with more than three times to spare, which is consistent and conservative. Transcend's 1,328, the only figure here that needed no assumption, reads the same way. SanDisk MAX Endurance at 5,484 and Samsung PRO Endurance at 6,406 are 1.8x and 2.1x above the 3,000-cycle ceiling SanDisk states for both MLC and TLC. Those two ratings are only consistent with a cell mode that outlasts ordinary MLC and TLC, and neither vendor names one anywhere on the product.

Rerun the arithmetic at what a dash cam actually writes

The rating is a Full HD number at 26 Mbps, and the datasheet says plainly that 4K gets fewer hours. It never says how many fewer, because that depends on your camera, so here is the same arithmetic run at the bitrates I have measured on this site's own dash cams. Hours scale inversely with bitrate: rated hours, times 26, divided by what the camera writes.

Two honest readings come out of that table, and they pull in opposite directions. On a single 4K channel, the rating's 26 Mbps is actually generous: the three single-channel 4K cameras I have measured average between 10.74 and 24.78 Mbps, all under the rating's assumption, so a 4K card in one of those is on pace for more hours than the box says, not fewer. Add channels and it flips. The busiest rig I have measured, a VIOFO A229 Plus recording two 2K channels, averages 53.46 Mbps, which cuts the High Endurance 128GB's 10,000 hours to about 4,863, a little over half a year of continuous recording. Its busiest single second hit 77.29 Mbps, and a peak is not a duty cycle, so I have named it and not projected it.

The card is not lying and the rating is not wrong. It is a single-channel Full HD number, correctly computed, printed on a product whose box also carries a 4K badge. What it never tells you is what your camera writes, and the storage calculator does, from the same measurements this table came from.

The warranty and the rating disagree

Both lines print a warranty on the same datasheet as the hours. Convert the hours into years of the continuous recording the card is sold for, 8,760 hours to the year, and set the two side by side.

Here is the text, so nobody has to take my word for the terms. The High Endurance datasheet's specification table reads "Warranty 2-Years," and the retail box in the corpus carries a roundel reading "2 YEAR / ANS LIMITED WARRANTY," labelled Card/Carte, card only. The MAX Endurance datasheet's table reads "256GB: 15 years, 128GB: 10 years, 64GB: 5 years, 32GB: 3 years," and the MAX 128GB box in the corpus prints its 10-year limited term in a circled mark, again labelled Card/Carte and footnoted card only.

Two things fall out. MAX Endurance's warranty scales with its capacity and tracks its own hours ladder at a steady 1.1x to 1.75x, which is a company treating the endurance figure as something it is prepared to stand behind in years. High Endurance's warranty is flat at two years against a ladder that quadruples from one end to the other, so it has to cross over: at 32GB the warranty is seven times longer than the rated life, at 128GB it is 1.75x longer, and at 256GB and above the rating outlives the warranty. For the 128GB card this bench has run, the datasheet says the card is good for 1.14 years of the recording it is sold for, and warrants it for two.

Three brands print no rating at all

Everything above is about the lines that publish a number. Five endurance-branded lines have been through my bench, and three of them do not.

• The Vantrue Endurance PRO 128GB reads "Endurance PRO" on the card face and prints no hours, no terabytes and no bitrate condition on the card or the box. It prints no speed figure either; the class marks are the entire testable surface.
• The VIOFO High Endurance line, at 32GB and 128GB and at 64GB, prints "High Endurance" on the box and "Industrial" on the card face, and no endurance rating on either, across all three capacities. No warranty term is printed anywhere.
• The Gigastone High Endurance Pro 128GB ships in a foil pouch with no product claim on it of any kind, and its product page states no hours and no terabytes. The card face does print "TLC," which is the only construction claim in the whole corpus and not one a bench can verify.

I want to be careful about what that is and is not. It is not evidence that those cards are worse; the Vantrue held 83.36 MB/s through its worst ten seconds on two readers, nearly the fastest of any endurance card I have run. It is a finding about disclosure: on those three lines there is nothing to audit, because the word is on the product and the number is not. SanDisk, whatever the arithmetic above says about its number, printed one, printed the condition behind it on the box, and printed a warranty next to it. That is the corpus benchmark for saying what you mean, and it is worth noticing which brands meet it.

What my bench can and cannot see

It cannot see lifespan. Nobody can, in a timeframe that would help a buyer, and the multi-day wear test that would come closest is one I have looked at hard and not run, because a meaningful version needs a dedicated rig, a plan for the thermal ceiling, and an honest answer to what a null result would mean. So there is no endurance number on this page from me, and there will not be one until there is a test behind it.

What a bench can see is how each of these cards behaves under the job they are sold for: five minutes of uninterrupted writing, with the worst ten seconds reported, and a whole-card fill read back byte for byte. Write speeds on this site are measured with every byte committed to the card, the way a dash cam writes, so they read lower than burst benchmarks; the scoring page explains why. Here is every endurance-branded card on the bench as of 19 September 2026, printed rating beside measured behavior, with the reader named on every number because the reader changes the answer.

Read that table for what it is. Every card held its fill and gave back every byte. Every card held the V30 floor it prints, and most held far more. The two SanDisk lines that print the biggest numbers are not the fastest writers on the table, and the Gigastone's two floors are a story about the state a card is in rather than about the card. None of those measurements is an endurance result, none of them contradicts an hours rating, and none of them confirms one. They are consistent with cards built for continuous recording, which is the most a bench can honestly say about a number nobody can test.

Reading an endurance rating in five steps

1. Find your capacity's own row. The headline is the biggest card's number. On SanDisk High Endurance a 32GB card is rated 2,500 hours, not 40,000.
2. Find the bitrate. It is on the back of the box, not in the datasheet, and it is 26 Mbps Full HD on every line I have seen that states one. Without it the hours are not a quantity.
3. Turn it into bytes. Hours times 3,600 times 3.25 is megabytes. 10,000 hours is 117 TB, which is 914 rewrites of a 128 GB card.
4. Rescale it to your camera. Rated hours times 26, divided by what your camera writes. The storage calculator has the measured bitrates.
5. Then read the warranty next to it. If the company scales the warranty with the rating, it is standing behind the number in years. If the warranty is flat, the rating is a comparison between capacities and nothing more.

What the number is for

The hours rating is good for one thing: comparing capacities within a line. It will correctly tell you that a 256GB card absorbs twice the footage of a 128GB card before either has been rewritten the same number of times, which you already knew. Across lines it tells you the ratio of two chosen constants, and the two biggest constants imply a cell mode nobody names. As a lifespan for your card, it is a Full HD single-channel projection with its condition printed on the packaging, and your camera decides how far off it is.

So read the capacity, read the bitrate on the box, read the warranty beside them, and then judge the card by what can be measured: whether it holds its speed for five minutes without a stall, whether every byte of its capacity comes back, and whether it does that on more than one reader. Those are the questions the reviews linked above answer, one card at a time. The hours are the capacity, multiplied, and the multiplication was done before the card was ever plugged in.

Sources

Every figure above comes from a manufacturer document, a manufacturer newsroom page, or the retail packaging of a card physically in the corpus. Nothing is quoted from a retailer listing. All four documents were fetched and text-extracted on 19 September 2026.

• SanDisk High Endurance microSD UHS-I card datasheet, Western Digital, rev 05/23: documents.sandisk.com. The hours ladder, footnote 1, the 2-year warranty.
• SanDisk MAX Endurance microSD UHS-I card datasheet, Western Digital, rev 01/2020: documents.sandisk.com. The hours ladder, footnote 1, the capacity-scaled warranty.
• Sandisk, Benefit of adopting Sandisk's high-endurance OEM-grade microSD cards, white paper, Matteo Zammattio, January 2026: documents.sandisk.com. The industrial P/E ceilings by cell mode and the two quoted sentences on consumer ratings.
• Samsung Semiconductor newsroom, Samsung Unveils New PRO Endurance Memory Card Optimized for Surveillance and Dashboard Cameras: semiconductor.samsung.com. The hours ladder and the 26 Mbps footnote.
• SanDisk High Endurance 128GB and MAX Endurance 128GB retail packaging, back panels, transcribed from the boxes of the units in the CKBench corpus. The 26 Mbps condition and the printed warranty roundels.
• Transcend High Endurance 350V terabytes-written figures: vendor product literature, not verified against a datasheet, and used here only as an aside.
• Dash cam bitrates: this site's own measurements, published at the dash cam storage calculator. Card measurements: the linked reviews, with every run on the CKBench hub.