PSS-10 norms: what the average person actually scores

The original norms

When Sheldon Cohen and Gail Williamson published normative data for the Perceived Stress Scale in 1988, drawn from a US probability sample of over two thousand adults, the mean PSS-10 score was about 13, with a standard deviation around 6. That single figure anchors everything about how the scale is read: a score near 13 is the middle of the distribution, a score near 19 is roughly one standard deviation above it, and a score of 25 or more is unusual — in the top few percent of that sample.

The spread is as important as the mean. Because the standard deviation is wide relative to the range, scores that look very different — 10 and 16, say — are both within the broad middle of the population. The scale is sensitive at the group level and useful for tracking one person over time; it is blunt for fine distinctions between individuals.

Rising over the decades

Cohen and Janicki-Deverts revisited the norms using US national samples collected in 2006 and 2009. The means had risen into the mid-teens — around 15 to 16 — with the increase visible across most demographic groups. Smaller studies since have reported means in similar or slightly higher territory, and samples collected during 2020–21 frequently reported means in the high teens, though those were rarely probability samples.

This matters when reading your score: the conventional bands below were anchored to the 1980s distribution. A score of 15 that would have been clearly above average in 1983 is close to the modern mean.

Figures are rounded; the original papers report subgroup means to one decimal place and should be consulted for research use.

Who scores higher

• Women score consistently higher than men by one to two points on average — a reliable finding across samples and decades, though small relative to the standard deviation.
• Younger adults score highest. Mean scores fall steadily with age; those over 65 typically score two to three points below those under 30. The decline is one of the most robust patterns in the data.
• Lower income and lower educational attainment are associated with higher scores, as is unemployment.
• People with a current diagnosis of depression or an anxiety disorder score much higher — often in the mid-to-high twenties. Perceived stress and these conditions overlap heavily, which is both why the PSS predicts health outcomes and why it is not a screening tool for them.

About the 14 and 27 cut-points

You will see the PSS-10 reported in three bands almost everywhere, including on this site: 0–13 low, 14–26 moderate, 27–40 high. It is worth being clear about what these are. They are not clinical thresholds and they did not come from Cohen. The scale's author has stated plainly that the PSS is not a diagnostic instrument and has no validated cut-offs. The bands are a reporting convention that emerged in applied research — roughly a split at the original mean and at about two standard deviations above it — and they have stuck because they are convenient.

Reading your score against the norms

A practical reading, using the modern mean of roughly 16 and a standard deviation of about 7:

• Below 10: lower perceived stress than about three-quarters of adults.
• 10–20: the broad middle of the population. Most people are here.
• 21–26: above about four in five adults. Worth paying attention to, especially if it is sustained.
• 27 and above: top few percent. In the range where structured support has better evidence than self-management, and where overlap with treatable conditions is substantial.

Remember that the PSS asks about the last month. A score taken the week after a bereavement, a move or a job loss is a valid measurement of a bad month, not of you.

Change over time in one person

The scale's test–retest reliability over a few days is high; over six weeks it drops, as it should, because perceived stress genuinely changes. That makes the PSS-10 better suited to tracking than to one-off judgement. A change of about five points or more between two administrations is unlikely to be measurement noise alone. If you use the Calmspan test to track, retake it at roughly monthly intervals rather than weekly — the instrument is defined over a month, and shorter gaps mostly measure the same month twice.