Average FPS vs 1 Percent Low: Which Number Tells You More?

Average FPS tells you how many frames a game produced per second over a whole recording; 1 percent low tells you how rough the slowest moments were. The first is a summary of speed, the second is a summary of consistency, and a game can score well on one while feeling poor because of the other. If you only look at one figure, look at the lows, because stutter is what players notice and complain about.

The short version

An average of 100 FPS with 1 percent lows near 85 FPS usually feels fluid. An average of 100 FPS with lows near 35 FPS usually feels uneven, even though the headline number looks great. The gap between the two is the story. A small gap means frames arrive at an even pace. A large gap means occasional frames take much longer than the rest, and your eye reads those as hitches.

That is why hardware comparisons that publish only an average can be misleading, and why this guide walks through how each number is built, what the gap means, and how to read both without over-reading either.

How average FPS is calculated

Average FPS is the simplest statistic in the toolbox: the number of frames rendered divided by the time the recording lasted. If a benchmark pass renders 6,000 frames in 60 seconds, the average is 100 FPS. It is easy to produce, easy to compare and widely used, and it answers a legitimate question: how much work can this system push through overall?

What it cannot do is describe how that work was distributed. Ten seconds of 200 FPS followed by a few seconds of near-freeze can still produce a healthy average, because fast stretches pay for the slow ones. The average is dominated by the typical frame and largely blind to the exceptional ones.

How 1 percent low is calculated

There are two common methods, and published numbers do not always say which was used. Method one sorts every frame by how long it took, selects the slowest 1 percent, and reports the average FPS of that group. Method two finds the 99th percentile frame time, meaning the time that 99 percent of frames beat, and converts it to FPS by dividing 1,000 by the milliseconds. Method one is generally a little lower and more sensitive to extreme spikes, while method two is more stable. Neither is wrong, but mixing them across charts invites bad conclusions.

A related figure, 0.1 percent low, applies the same logic to the slowest tenth of one percent. It highlights rare, severe hitches, but it rests on very few frames, so it can swing widely between runs unless the recording is long.

A worked example with frame times

Numbers make the idea concrete. Imagine a short capture of 1,000 frames under these stated assumptions: 950 frames take 8 ms each, 40 frames take 14 ms each, and 10 frames take 60 ms each. These are invented figures chosen for clear arithmetic, not data from any real game or system.

Average FPS is 1,000 frames divided by 8.76 seconds, which is about 114 FPS. Now the 1 percent low by method one: the slowest 1 percent of 1,000 frames is 10 frames, and those are the 60 ms spikes. Their average is 1,000 divided by 60, about 17 FPS. By method two, the 99th percentile sits at the border between the medium frames and the spikes; if it lands on the 14 ms group, it converts to roughly 71 FPS, and the exact value depends on how the tool handles ties. The takeaway is that the same capture can yield 114 FPS average and a low somewhere between 17 and 71, depending on the method, which is exactly why you need to know which one a source uses.

This capture would feel mostly smooth with ten visible hitches, and that matches how people describe games with good averages and bad lows.

Side-by-side comparison

What a wide gap usually points to

When lows sit far below the average, the cause is usually a resource that occasionally runs short. Video memory is a frequent suspect: once a game overflows the card’s pool, data moves over the PCIe slot and frames stall. A CPU that cannot prepare frames quickly enough can cause periodic dips in busy scenes, especially in games with many characters or physics objects. Slow storage can add stutter when a game streams assets while you move. Shader compilation, background updates, overlay software, overheating and capture tools running at the same time can all add spikes of their own.

The useful habit is to change one thing at a time and re-record. If lowering texture quality lifts the lows while the average barely moves, memory was the limit. If lowering resolution raises the average but not the lows, the CPU or storage may be the bottleneck. Our walkthrough on fixing low FPS covers the broader checklist for tracking down slowdowns.

How your display changes the answer

On a fixed 60Hz screen with vsync on, a missed refresh shows up as a repeated frame and a visible judder, so a dip below 60 is plainly felt. With a variable refresh rate display, the screen follows the GPU, so moderate dips are much less obvious, though large swings still read as uneven motion. A higher refresh rate also tightens the target: at 144Hz, the frame budget is about 6.9 ms, so a spike to 20 ms is more noticeable than it would be at 60Hz. Cable and port choices can limit which refresh rates are available at all, which our comparison of DisplayPort and HDMI 2.1 explains in detail.

What to look for when reading benchmarks

• The method behind the lows. Look for a note on whether lows are an average of the slowest slice or a percentile conversion.
• The recording length. Very short runs make 1% and especially 0.1% figures unreliable.
• The scene. A fixed in-game benchmark and a free-roaming run produce different results, and neither is wrong, but they should not share a chart.
• Frame-time graphs. A graph shows the spikes directly, and one glance often tells more than a table.
• Matching settings. Resolution, texture quality, upscaling mode and ray tracing must match across systems being compared.
• Other software running. Overlays and capture tools can lower both numbers.

Common mistakes

The biggest is comparing an average from one source with a low from another, which mixes methods and scenes. The second is treating a 0.1 percent low from a one-minute run as a verdict, when a single level load can drag it down. The third is chasing the average with an upgrade when the real problem is the lows; a faster GPU will not fix stutter caused by slow storage or too little memory. The fourth is ignoring your own screen, since a variable refresh display and a fixed one react to the same dips differently. The fifth is judging by feel in one session and by numbers in the next, so pair them: record a run while noting when you felt hitches, then check whether the dips line up.

When the numbers are not enough

If lows are close to the average but the game still feels wrong, look at input lag, frame pacing under vsync or a mismatch between the frame cap and the refresh rate. A cap a few frames below the refresh rate often smooths pacing on a fixed-refresh screen. If the lows are bad and stay bad across settings, record a longer run, log temperatures and clock speeds, and check whether the card or processor is throttling before spending money on parts.

Related guides

• Logitech G Pro X 2 Vs Steelseries Arctis Nova Pro Wireless
• Travel Controller Case For 8Bitdo Ultimate Series
• Gaming Pc Under 600 For Roblox And Minecraft

Frequently asked questions

What does 1 percent low FPS actually mean?

It describes the slowest slice of a recording. The most common method sorts all frame times, takes the slowest 1 percent, and reports the average frame rate of that slice. Another method reports the 99th percentile frame time converted to FPS. Both aim to capture how bad the worst moments felt, which an average smooths away. Tools differ, so compare numbers produced by the same method.

What is a good gap between average FPS and 1 percent lows?

There is no official threshold, but a rough guide is that lows within about 70 to 80 percent of the average usually feel smooth, while lows below half the average often feel choppy. The right target also depends on your display. On a variable refresh rate monitor, a modest gap is more forgiving than on a fixed 60Hz screen with vsync on.

What are 0.1 percent lows?

The same idea applied to an even smaller slice: the slowest one tenth of one percent of frames, sometimes called the worst-case spikes. Because it samples so few frames, it is sensitive to a single hitch such as a level load or shader compile. It is useful for spotting rare stutter, but a short recording can make it noisy, so longer runs give more trustworthy figures.

Can a game feel smooth with a low average FPS?

Yes. A steady 60 FPS with very even frame times can feel better than 90 FPS that swings between 40 and 140. Consistency matters because your eyes notice sudden changes in motion cadence. That said, a low average does limit how responsive the game feels, so the ideal is a high average with lows that stay close behind it.

What causes poor 1 percent lows?

Common causes include running out of video memory, a CPU that cannot keep up with draw calls, slow storage during asset streaming, background applications, shader compilation, thermal throttling and an overloaded capture setup. Because many causes are possible, change one variable at a time and compare lows before and after rather than guessing from the average alone.