What is a normal digit span?
The honest answer is that there is no single number — and the reason is more useful than a number would have been. "Digit span" is not one test. It is a family of procedures, and the procedure you used sets the figure you get.
The short answer
Established
The clearest illustration comes from a 2011 study by Woods and colleagues, which tested 763 adults aged 18 to 65 on forward digit span and scored the very same sessions three different ways.
| How it was scored | Mean span |
|---|---|
| Longest length reached before failing twice at the same length | 6.35 digits |
| Mean span across all trials | 6.52 digits |
| Longest list correctly recalled at any point | 6.77 digits |
Same people. Same sitting. Three different answers. Nothing about anyone's memory changed between those three figures — only the rule for turning their performance into a number.
So if you have arrived here holding a number and wanting to know whether it is normal, the first thing worth knowing is that the published figures you will find elsewhere disagree with each other for reasons that have nothing to do with the people being tested.
Why one test gives three numbers
Established
Digit span looks simple: you hear or see a sequence, you give it back. The complexity is entirely in the scoring, and there are at least three defensible rules.
- Stop after two failures at the same length. The traditional clinical approach. It is quick, but it ends the test at the first patch of bad luck — an unusually awkward sequence, or a moment of inattention, ends the measurement early.
- Take the longest list ever recalled correctly. More generous, and it captures what someone can do on a good trial rather than where they first stumbled.
- Take a mean across many trials. Produces a figure with decimal places rather than whole digits, and is less thrown by any single trial.
In that study the first rule underestimated the third by roughly half a digit on average. Half a digit sounds trivial. It is not, when the range is this narrow: under the same rule, 80% of those 763 adults scored between 5 and 7 digits, with a median of 6.
Who you test changes it as much as how you score it
Established
The same researchers ran a smaller experiment with 30 adults, average age 26, with an average of 14.8 years of education. Scored by the two-error rule — the same rule that gave 6.35 in the larger group — this group averaged 7.36 digits.
The larger group of 763 had an average age of 46.5 and an average of 12.5 years of education. That is a full digit of difference between two samples measured by the same team, using the same equipment, with the same scoring rule.
Within the 763-person sample, forward span did not show a statistically significant relationship with age. Backward span did, and it was small and negative. Both forward and backward span were positively related to years of education, and the relationships were weak in every case — these are group tendencies, not a formula you can apply to yourself.
Forward and backward are not the same task
Established
In the 763-person sample, mean forward span was 6.35 digits and mean backward span was 4.61 digits, scored the same way. Recalling in reverse is reliably harder, and the research treats forward and backward span as measuring partly different things rather than one ability at two difficulty settings.
This matters when comparing numbers you find online: a figure quoted without saying whether it is forward or backward is missing the single largest piece of context it could carry.
What about "seven, plus or minus two"?
Emerging and contested — read this section as an active disagreement, not a settled fact
Almost everyone arrives at this question having heard that the answer is seven. That figure traces to a 1956 paper by George Miller, and it has had a strange afterlife.
Cowan, reviewing the evidence in 2001, characterised Miller's seven as having been intended more as a rough estimate and a rhetorical device than as a genuine capacity limit. Cowan argued instead for a central limit averaging about four chunks, noting that others had proposed three to five.
We are flagging this as contested rather than settled, and the sources we cite are the reason: seven, four, and three-to-five are all in them. The figure you get also depends on whether the material can be grouped into larger units — holding 1 9 4 5 is not the same task as holding four unrelated digits if 1945 means something to you.
The practical consequence is narrow but real: a capacity limit measured in chunks and a digit span measured in digits are not the same quantity, and treating "seven" as the target your span should hit is comparing two different things.
Why your score from our test is not one of these numbers
Established, and this is the most important section on this page
Every figure above comes from a test that was administered by a trained examiner, presented as recorded audio through headphones, and answered by speaking aloud, with the examiner transcribing.
Our test shows digits on a screen and takes your answer on a keypad, with nobody in the room. Those are different tasks with the same name.
This is not our caveat — it is the authors'. They note that differences between presentation methods are significant, particularly for backward span, and that self-administered computerised tests introduce variation from motivation, from how well the instructions were understood, and from comfort with the device, none of which is present when an examiner runs the test.
So a number from our test cannot be checked against the figures on this page. It is not that the comparison is discouraged; it is that the two numbers are measurements of different procedures.
We have not measured anyone. Our reference sample is zero people. We publish that figure as a zero rather than omitting it.
That is why our test gives you a number and an explanation of what it does and does not mean, and does not tell you where you sit relative to other people. We have no basis for saying that, and we are not going to imply one. The methodology page sets out exactly where we are and what would have to change.
Take the digit span test → — about two minutes, no account, nothing stored.
What it would actually take to answer this properly
Established
A defensible statement about typical performance on a specific instrument needs, at minimum: a sample large enough to be stable, a description of who is in it and how they were recruited, a fixed and published scoring rule, published exclusion rules applied before anyone looks at the results, and a stated presentation and response method.
The studies on this page have all of that. That is why their numbers are quotable — and also why their numbers belong to their procedures rather than to digit span in general.
Anyone offering you a single clean figure for "normal digit span" without naming the scoring rule, the sample and the administration method is not simplifying the answer. They are omitting the parts that determine it.
If you came here because you are worried about your own memory, an online test cannot answer that question and neither can an article. A doctor can. Nothing here can tell you whether something is medically wrong, and a low score on any test on this site does not indicate a condition.
What this page does not tell you
Open — stated because leaving it out would make the page look more certain than the evidence is
- Whether your own score is typical. We cannot tell you that, for our test or anyone else's, and no figure on this page should be read as a threshold you passed or failed.
- What one measurement is worth. A single session, on one day, in one set of conditions, is a noisy estimate of anything.
- How the figures here would change with a different population. Both samples described above were specific groups of adults, recruited for specific studies. Neither was assembled to represent everyone.
Sources
- Woods, D. L., Kishiyama, M. M., Yund, E. W., Herron, T. J., Edwards, B., Poliva, O., Hink, R. F., & Reed, B. (2011). Improving digit span assessment of short-term verbal memory. Journal of Clinical and Experimental Neuropsychology, 33(1), 101–111. doi:10.1080/13803395.2010.493149
- Cowan, N. (2001). The magical number 4 in short-term memory: a reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87–114. doi:10.1017/S0140525X01003922