What Is 1% Low FPS? The Number That Predicts Stutter

Two systems can average the same frame rate and feel completely different. The statistic that separates them is the 1% low — the frame rate of the slowest one percent of frames — and once you understand it, average FPS never quite convinces you again.

Check your own setup first: the stutter test measures your display live, in this browser, in about five seconds. Run it now

Why the average lies

An average is a smoothing machine: it takes a thousand frames, most of them fast and a few of them dreadful, and reports the comfortable middle. But your eyes do not experience the middle — they experience the sequence, and the dreadful frames are the ones you notice, because a single 50-millisecond frame in a stream of 8-millisecond frames is a visible hitch, a moment where the world pauses. A system averaging 120 FPS with dips to 35 feels worse in those moments than a system locked at a steady 80, and any honest smoothness metric has to price the dips. That is the entire argument for the 1% low.

The 0.1% low extends the same logic to the rarest events — the one-in-a-thousand frame that marks a level load, a shader compiling, a background process waking up. Reviewers quote both because they answer different questions: the 1% low describes how the game feels minute to minute; the 0.1% low describes its worst interruptions.

How it is calculated

Start from frame times, not frame rates. Record every frame's duration during a run, sort the list slowest-first, take the slowest one percent, and average their durations. Convert that average back — 1,000 divided by the milliseconds — and you have the 1% low FPS. The browser test on this site performs exactly that computation live, which is why its lows are measurements rather than estimates. The calculator pages use a different, clearly-labelled approach: a per-game ratio between lows and average drawn from published reviews, because a calculator has no live frames to sort. Knowing which kind of number you are looking at — measured or modelled — is part of reading either honestly.

A worked example makes the gap vivid. Suppose 990 frames take 8 ms and 10 frames take 48 ms. The average frame time is about 8.4 ms — a headline of roughly 119 FPS. The 1% low averages those ten 48 ms frames: about 21 FPS. Both numbers are true. Only one of them describes what you felt during those ten frames.

What good looks like, and how to improve yours

As a working rule, a 1% low within about ten percent of the average means genuinely even delivery; within twenty percent is normal for demanding games; beyond that, investigate. The causes mirror the stutter checklist because they are the same phenomenon seen statistically: background load, thermal throttling, memory pressure, asset streaming and shader compilation. Fixes follow the same order — close the background, cool the machine, feed it memory bandwidth, and only then consider hardware. And when comparing systems or settings, compare the lows first: a change that raises the average by ten but the lows by twenty is the better change, every time, for how the game will actually feel.

Now compare what you found: run the stutter test again after any change — the number that moves names the bottleneck.

Frequently asked questions

What is the difference between 1% low and minimum FPS?

Minimum FPS is a single frame — the worst one — and is easily distorted by one loading hitch. The 1% low averages the slowest one percent, making it stable enough to compare across runs and systems.

What is a good 1% low for 144Hz gaming?

Ideally within about 10% of your average — so near 130 if you average 144. In demanding AAA titles, lows around 100 with a 144 average are common and still feel fine; the ratio matters more than the absolute gap.

Why do reviews quote 1% lows so much?

Because they correlate with perceived smoothness far better than averages. Two cards with equal averages and different lows feel different, and the lows are what players describe as stutter.

Can I improve 1% lows without new hardware?

Often, yes: background processes, thermals, single-channel memory and shader-compilation behaviour all hit the lows hardest, and all are fixable or manageable in software. The stutter test helps identify which one you have.

Related tools

FPS TestRefresh Rate TestMonitor Test HubFPS Calculator