The Trail Making Test Part B times how quickly you can connect an alternating number and letter sequence. It measures processing speed and cognitive flexibility rather than knowledge or intelligence, which matters because those are the abilities that decline earliest and respond most to sleep and cardiovascular fitness.
- There is no single normative time. Means run from about 49 seconds at ages 18–24 to about 131 seconds at 75–79, stratified by education as well as age.
- The 3-or-3 rule, three minutes or three errors, is a fitness-to-drive screen rather than a cognitive impairment threshold.
- Longevity Coach IQ tier boundaries are an estimate scaled from that average, not a published percentile table.
- Sleep and aerobic exercise have the strongest evidence of anything for processing speed.
What the Trail Making Test actually measures
The task is simple to describe: connect 1 to A, A to 2, 2 to B, and so on to 13, as fast as you can without lifting the pen or making errors. What it measures is less obvious.
Two abilities are being timed at once. Processing speed is how quickly you take in and act on information, and it is the cognitive ability that declines earliest, beginning in the twenties. Cognitive flexibility is the capacity to hold two sequences in mind and switch between them, which is an executive function.
What it doesn’t measure is worth stating: not knowledge, not vocabulary, not reasoning, not intelligence. Those are largely crystallised abilities, and they follow a completely different trajectory with age, covered in cognitive aging.
How a Trail Making Test time is scored
Lower is better, and the reference points are:
| Age | Education | Mean TMT-B (s) | SD |
|---|---|---|---|
| 18–24 | — | 48.97 | 12.69 |
| 45–54 | 12+ years | 58.46 | 16.41 |
| 55–59 | 0–12 years | 78.84 | 19.09 |
| 55–59 | 12+ years | 68.74 | 21.02 |
| 60–64 | 0–12 years | 74.55 | 19.55 |
| 60–64 | 12+ years | 64.58 | 18.59 |
| 65–69 | 0–12 years | 91.32 | 28.89 |
| 70–74 | 12+ years | 109.95 | 35.15 |
| 75–79 | 12+ years | 130.61 | 45.74 |
| 85–89 | — | 132.15 | 42.95 |
Read your time against your own age and education band, not against a single average. A 75-year-old at 130 seconds is at the mean for their band. Against an "average of 75 seconds" the same person looks impaired, and isn't.
Impairment is conventionally defined as 1.0 to 1.5 standard deviations below the age- and education-corrected mean, which is a different number of seconds for everyone. That is why no fixed concern threshold appears here.
The "three minutes or three errors" rule you will see quoted is a fitness-to-drive screening convention, not a cognitive impairment cut-off. Tombaugh found the mean under 180 seconds for every age group.
Education shifted TMT-B results by up to 30 seconds in German population norms, and performance declined by about 3 seconds per year of life.
A time can only be read against your own age and education band. The same number means different things thirty years apart: 100 seconds is unremarkable at 70 to 74 and unusually slow at 18 to 24.
An honesty note on the tiers
Most markers in this encyclopedia are scored against published percentile data for your age and sex. This one isn’t, and the distinction matters. Because the tiers are scaled from a single average rather than from the age- and education-stratified distribution, they will systematically mis-rank older users and users with fewer years of education, which is a further reason to read this marker as a personal trend rather than a tier.
Tombaugh's normative data is real and well established, but it comes from the pen-and-paper version administered in person. A self-administered digital version isn’t the same test: screen size, input method, familiarity with touch interfaces and the absence of an examiner all shift the times. No published percentile table exists for it.
So the tier boundaries used here, roughly 45, 55 and 65 seconds for the top bands, are an estimate scaled from the published average rather than a measured distribution. That is a weaker basis than VO2 Max or grip strength have, and it is flagged rather than glossed over. Track your own trend and treat the tier as indicative.
What happens if you cannot finish the test
In a Japanese community sample of 1,966 older adults, 82.6% completed TMT-B within the 300-second limit.
In a German population sample, 10.9% of participants over 74 did not complete within 300 seconds, rising to 13.9% among those over 74 with the lowest educational level.
A non-completion is a data point rather than a failure state. That matters practically for a self-administered digital version, where a user who cannot finish needs to be told what it means.
Why Part B and not Part A
Part A connects the numbers 1 to 25 in order. It is a clean measure of visual scanning and motor speed, and it is largely preserved until quite late in cognitive decline.
Part B adds the alternation, and with it the executive demand of holding two sequences and switching between them under time pressure. That switching cost is what makes Part B considerably more sensitive to the early changes worth detecting.
In clinical use the difference between the two times, B minus A, is sometimes used to isolate the switching component from raw speed. If you have both numbers, that difference is the more interesting one.
What actually moves it
- Sleep. The strongest and fastest-acting lever. Processing speed is measurably impaired by sleep restriction. See sleep duration.
- Aerobic exercise. The best-evidenced lifestyle intervention for cognition generally. The mechanism likely runs partly through vascular health, which is why cardiovascular and cognitive risk factors overlap so heavily.
- Sustained mental engagement. Learning a language or instrument, or strategy games. Real but less direct than the first two, and subject to the narrow-transfer problem: you get better at what you practice.
The practice effect is worth naming as a limitation of the test rather than a lever. Getting faster at the Trail Making Test isn’t the same as getting faster generally.
The dementia connection
This is one of the markers feeding the Cognitive Decline and Dementia risk category, alongside the modifiable risk factors and homocysteine.
Its value there is as a functional baseline. A single reading says little, but a time recorded in your forties gives you something to compare against in your sixties, and a trajectory is far more informative than any one-off measurement. Most people never establish that baseline, and can’t construct it retrospectively.
The honest framing: this test doesn’t predict dementia. It measures an ability that declines early, which makes it useful for noticing change while there is still time to act on the modifiable risks.
Reading it as a trend, not a score
Cognition sits inside Functional Capacity, which is deliberate. Staying mentally quick and staying physically capable are usually treated as separate projects, and they share more machinery than that framing suggests: vascular health, sleep quality and aerobic fitness feed both.
Because the tier boundaries here rest on an estimate rather than a percentile table, this marker is best read as a trend line rather than a score. A stable time across years is the signal worth having.
Sources
Key references for the claims on this page. Where a figure is attributed to a specific study or body, it is named here.
- Tombaugh TN. Trail Making Test A and B: normative data stratified by age and education. Archives of Clinical Neuropsychology, 2004. The age- and education-stratified normative dataset used in the table above. Archives of Clinical Neuropsychology, 2004;19(2):203–214. PMID 14732469
- The 3-or-3 rule, three minutes or three errors, is a commonly cited clinical screening convention rather than a finding from a single study.
- Honesty note: the tier boundaries used here, roughly 45, 55 and 65 seconds for the top bands, are Longevity Coach IQ's own estimate scaled from Tombaugh's average. No published percentile table exists for a self-administered digital version of this test, and presenting one as though it did would be inventing precision.
- Salthouse TA. When does age-related cognitive decline begin? Neurobiology of Aging, 2009;30(4):507–514. Processing speed as the earliest-declining cognitive ability. DOI
Frequently asked
What is a good Trail Making Test B time?
It depends entirely on your age and education, and a single average is misleading. Tombaugh’s norms put the mean at about 49 seconds for ages 18–24 and about 131 seconds at 75–79. Read your result against your own band. Impairment is conventionally 1.0 to 1.5 standard deviations below that corrected mean, not a fixed number of seconds. The three-minute figure often quoted comes from fitness-to-drive screening.
Does a slow time mean I have dementia?
No. This is a screening measure of processing speed, not a diagnostic test, and a single slow attempt tells you very little. Tiredness, unfamiliarity with the format, a small screen and distraction all add seconds. A consistently slow time across attempts, or a clear decline over years, is what would be worth raising with a doctor.
Why does the test alternate numbers and letters?
That alternation is the entire point. Part A connects numbers alone and mostly measures visual scanning and motor speed. Part B forces you to switch between two sequences, which adds the cognitive flexibility component. The difference between your two times isolates the switching cost.
Can I practice to get better at it?
You will get faster at this specific task, which is a practice effect rather than improved cognition. That is why a first attempt is often the most informative one, and why comparing your time against your own history is more useful than comparing against a table.
What actually improves processing speed?
Sleep and aerobic exercise have the strongest evidence. Consistent 7 to 9 hour sleep and regular cardiovascular training both show effects on processing speed specifically. Mentally engaging activity helps, though the evidence there is less direct.
Do brain training apps help?
You get better at the games. Transfer to unrelated cognitive tasks is where the evidence weakens considerably, and that is the finding across brain-training research generally. Doing them because you enjoy them is fine; doing them as insurance is harder to justify.
How often should I retest?
Every three to six months at most. Processing speed doesn’t change quickly, and frequent retesting mainly measures practice and how rested you are on the day.
Is this the same as an IQ test?
No. It measures speed of processing and mental flexibility, not knowledge, reasoning or vocabulary. Someone with high crystallised intelligence can post an ordinary time, and someone quick on this task isn’t thereby more intelligent.