Attention span test (sustained attention)

How to take the test

1. Under Length of the test, choose Full (225 numbers, 4 minutes 19 seconds) or Short (117 numbers, 2 minutes 15 seconds).
2. Press Start the practice. It runs 18 numbers in about 20 seconds and shows a hint under the display after each mistake.
3. Press Start the test. The numbers 1 to 9 appear one at a time in the box, each followed by a circle with a cross. Press Space, or tap the big button, for every number except 3. When a 3 appears, hold back.
4. Keep this tab in front until the end. Switching tabs or apps stops the run without saving it, and so does pressing Escape or Stop.

Your results appear as soon as the last number has gone, with your earlier runs on this device underneath.

How your result is worked out

The task is the Sustained Attention to Response Task (SART), introduced by Robertson and colleagues in 1997. It asks you to withhold a press on rare targets, here the 3, which is one number in nine. The settings follow the Millisecond SART manual: each number shows for 250 milliseconds, then the mask for 900, so a new number arrives every 1.15 seconds. The full run shows each digit 25 times in random order, 225 in all, in five font sizes used 45 times each.

From your presses, the result works out:

• 3s held back: the share of the 25 threes you didn’t press on. This is the big number.
• Presses on a 3: the slips, as a count and a percentage.
• Missed presses: other numbers you didn’t press for, out of 200.
• Average response time: from the moment a number appears to your press, over correct presses of 200 milliseconds or more. As in the manual, presses under 100 milliseconds count as anticipations, and those from 100 to 199 are left out of the times.
• Variation: the standard deviation of those times, and the coefficient of variation (CV), which is the standard deviation divided by the average.
• Before your slips: the average of the 4 correct presses before each press on a 3, set against your average for the whole run.

Here’s a worked example. Say you press on 2 of the 25 threes, miss 1 of the 200 other numbers, and your correct presses average 420 milliseconds with a standard deviation of 63. The result shows 92% of the 3s held back, 8% presses on a 3, 0.5% missed presses and a CV of 0.15 (63 divided by 420).

The page times the display with your browser’s animation frames and logs how long each number really stayed on screen. If a run is interrupted, a frame stalls long enough to skip a number, or the numbers’ average display time drifts more than 10% from 250 milliseconds, the run isn’t saved, because its times wouldn’t be comparable with your others.

The 8-second figure and the goldfish

A figure that does come from a study is 47 seconds. In a study published in 2016, Gloria Mark and colleagues logged 40 office workers’ computers for two weeks. People stayed on one screen for 47 seconds on average before switching, with a median of about 40. That’s a count of window switching at work, which is a different thing from how long a person can concentrate.

The claim that lectures lose students after 10 minutes has a similar history. In a 2016 article reviewing the evidence, Bradbury concluded that the available primary data don’t support a 10 to 15 minute attention limit.

What this test measures

This test measures sustained attention: keeping a simple, repetitive response going for minutes while staying ready to stop it. With 8 numbers in 9 needing a press, pressing can settle into a rhythm. Robertson and colleagues found that false presses on the SART could be predicted by a shortening of response times on the trials just before, which is why the result compares your last 4 presses before each slip with your average.

So the result is a set of counts and times from one run of a few minutes. It doesn’t give a number of seconds you can focus for, and no research gives a figure like that for anyone.

How to read your results

Read the presses on 3 and the missed presses together. Holding back on every 3 while missing many other numbers can mean you slowed down to play safe; pressing fast with few misses but several slips is the opposite trade. Your average response time and its CV show how evenly you responded: a CV of 0.15 means your times typically varied by about 15% of your average.

Age makes a difference too. In a study of 638 people aged 14 to 77, Carriere and colleagues found that SART errors and response speed both declined in a straight line with age, while response time variability dropped early in life and then stayed level. That study gives no cut-off scores, so this page doesn’t compare you with other people. It compares you with your own earlier runs, kept in this browser only.

Use the same device each time. Keyboards, touch screens and displays each add their own small delay between your press and the moment the page records it, so a phone and a laptop can give different times for the same person.

Phones and focus

The research here is mixed, and most of it is correlational. A 2025 review by Nguyen and colleagues pooled 71 studies with nearly 100,000 people and found that heavier short-form video use was associated with weaker attention (a correlation of about minus 0.38) and weaker self-control. Most of those studies measured both at once, so they can’t say which came first.

On the phone sitting next to you, the findings disagree. Ward and colleagues reported in 2017 that having your own phone nearby reduced available cognitive capacity, and a preregistered replication by Ruiz Pardo and Minda in 2022 found that the effect didn’t replicate.

What the result doesn’t tell you

One run is a small sample: with only 25 threes, each slip moves your percentage by 4 points, so one slip more or less means little. If you want to see whether a change makes a difference to you, run the test several times before and after, at the same time of day and on the same device.