CPT online: continuous performance test, A then X

Letters appear one at a time, every 2.5 seconds. You press only when an X comes straight after an A and let every other letter pass. Over 300 letters, the result shows how well you caught the targets and how even your responses stayed.

Start the CPT

300 letters · about 15 min · space bar or tap

When the last letter has gone by you see your result and can keep it with a free account.

K A X B X
A, then X: pressX after B · let it pass

How it works

A 20-letter practice with feedback comes first, then 300 scored letters with one pause halfway. With the instructions it takes about 15 minutes.

  1. 1 · Practice

    20 letters with feedback

    After each letter, a short line says whether it was a hit, a miss, a false alarm or a correct pass. Practice is not scored, and a 3-second countdown follows it.

  2. 2 · Target

    X straight after A

    Press the space bar, or tap the screen on a phone, when an X directly follows an A. That pair comes up 60 times in the 300 letters.

  3. 3 · No press

    Everything else

    Let all other letters pass: an A followed by another letter, an X after any letter but A, and B, C, D, E, H, K and L. A press on the A itself is also a false alarm.

  4. 4 · Pace

    A letter every 2.5 seconds

    Each letter is on screen for half a second, and a press counts up to 1.5 seconds after it appears. After 150 letters a pause screen waits until you choose to carry on.

No feedback appears during the scored letters, and a press does not speed up or slow down the stream. The order is shuffled on every attempt: 60 A-then-X pairs, 30 pairs where A is followed by another letter, 30 X's with no A before them and 90 other letters.

Continuous performance tests are a family of attention tasks with many versions, and Riccio and colleagues (2002) review what they show about the brain systems behind attention. This one uses the A-then-X rule, so every letter also asks you to keep the one before it in mind.

What it measures

Each single letter is easy. What the task tests is whether you can keep it up, accurately and at an even pace, for 300 letters.

Catching targets

  • Sustained attention: staying with a slow, repetitive stream long enough to catch each A-then-X pair.
  • Holding the previous letter in mind, because an X only counts when an A came just before it.
  • Your speed on the targets you catch, in milliseconds.

Holding back

  • Response control: not pressing for an X that follows another letter, or for whatever follows an A.
  • Steadiness: how much your response time changes from one target to the next.
  • Staying alert through long runs of letters where nothing calls for a press.

Missing a target and pressing at the wrong moment are different slips, so the result keeps them apart. The data panel even splits wrong presses by type: an X without an A, a letter after an A, and any other letter.

Both kinds of measure are well studied. Meta-analyses have pooled CPT results across ADHD studies (Huang-Pollock et al., 2012), and one review gathered 319 studies of reaction time variability in ADHD (Kofler et al., 2013). Yet in a clinic sample of 69 adults referred for ADHD assessment, scores from one commercial CPT did not separate those who were diagnosed from those who were not (Adamou et al., 2022), and in children results from another CPT correlated only weakly to moderately with parent and teacher ratings (Johnsen et al., 2024). A CPT result is one piece of information, never an answer on its own.

A test, not a diagnosis.

This is one run of one long task. On its own it cannot confirm or rule out any condition.

What you get

Your result appears the moment the 300th letter has gone by.

On the result screen

  • Your d-prime, your response time variability and your hit rate, at the top.
  • A summary line and a marker on a low-to-strong scale, both from fixed d-prime bands: below 2, 2 to 3, and 3 or more.
  • Notes on your hit rate, your false alarm rate, your variability and, when it is above 15%, the share of targets you missed.
  • A data panel with mean response time and its spread, each rate with its counts, the wrong presses by type, and a CSV of every scored letter.

When you keep it

  • Kept in your results once you create a free account, on any device you sign in from.
  • Shown under attention beside the other attention tests you take.
  • Followed by a suggested next test, so more areas fill in.

How this version is scored

Every scored letter ends as a hit, a miss, a false alarm or a correct pass. The test counts them and builds each figure from the counts.

Hit and miss
A hit is a press on an X that came right after an A. A miss is one of those X's left without a press. Only your first press on a letter counts.
False alarm
A press on any of the other 240 letters, the A's included. Letting one pass is a correct rejection.
d-prime (d′)
How well your presses separate targets from everything else: d′ = z(hit rate) − z(false alarm rate). Both rates are first held between 1% and 99%, so even a perfect run gives a finite value, at most about 4.65.
Response time and variability
Mean response time is the average over hits only. Variability is the coefficient of variation: CV = standard deviation of hit times ÷ mean hit time × 100. A higher CV means less even responding.
Accuracy
Hits plus correct rejections as a share of all 300 letters. Since 80% of the letters need no press, accuracy stays high even when many targets are missed, which is why d′ leads.

What the bands are. The summary line and the scale use fixed d′ cut-offs, and each note uses fixed cut-offs on its own rate. They are the same at every age and are not percentiles from a norm group.

Not the same as other CPTs. Commercial CPTs differ in letters, pace, target rule and length, and many report scores against their own norms. Figures from those tests do not carry over to this one.

Before you start

300 letters is a long stretch, so set aside about 15 quiet minutes.

For a steadier run

  • On a computer, rest a finger on the space bar. On a phone, set it down and keep a thumb close to the screen.
  • Silence notifications and stay on the tab: leaving it partway through breaks the stream.
  • Use the pause at 150 letters to stretch or rest your eyes, then go on.
  • Keep the rule simple: watch for an A, then be ready for an X.

Interpret cautiously when

  • You were tired, short of sleep or unwell; short-term sleep loss affects attention tasks (Lim and Dinges, 2010).
  • Something interrupted the run, or you stepped away at a point other than the halfway pause.
  • Your device lags or registers taps late, which shifts response times.
  • You are used to letter tasks like this one or took this test recently.

Questions

What does a continuous performance test measure?

Mainly sustained attention: how well you keep catching rare targets in a slow, repetitive stream. This version also asks you to hold the previous letter in mind and to hold back on near-targets, such as an X that does not follow an A. The result separates missed targets from wrong presses and adds how steady your response times were.

What is a good CPT score?

There is no single cut-off, because CPT versions differ in letters, pace, rules and length. On this one, the result screen places a d′ of 3 or more at the strong end and below 2 at the low end, using fixed bands, while a d′ near 0 would mean your presses did not follow the targets at all. Compare your own attempts made under similar conditions rather than results from other CPTs.

Can I take the CPT on my phone?

Yes. On a phone you tap anywhere on the screen instead of pressing the space bar, and a hint at the bottom reminds you. Touch screens can add delay to response times, so set the phone down, keep it steady and compare phone attempts with phone attempts.

Why is the CPT so long?

Because keeping focus over time is what it measures. A few minutes of letters is easy for most people; the task looks at whether hits, wrong presses and response times hold up across 300 letters. The pause at the halfway point is the only planned break.

Sources

By Cristian Dominguez Rein-Loring, Creative Technology Lead · Updated October 12, 2026