DASS-21 Scoring and Interpretation: What the Depression, Anxiety, and Stress Numbers Mean

Peter Lovibond and Syd Lovibond built the Depression Anxiety Stress Scales at the University of New South Wales, publishing the 42-item original in a 1995 manual and the supporting validation work the same year in Behaviour Research and Therapy. The problem they set out to solve is one every clinician has run into. Depression and anxiety self-report measures correlate so heavily with each other that two elevated scores often tell you almost nothing about which construct is driving either one.

Their answer was to give the shared component its own scale. What the Lovibonds called Stress covers persistent tension, difficulty relaxing, irritability, and a short fuse: the arousal that shows up in depressed clients and anxious clients alike. Pulling it out left the Depression scale free to concentrate on dysphoria and hopelessness, and the Anxiety scale free to concentrate on physiological arousal and situational fear.

The DASS-21 is the short form of that instrument. Seven items per scale instead of fourteen, the same response format, the same past-week reference window. Antony and colleagues (1998) published the first psychometric data on the short version and found it produced a cleaner factor solution than the 42-item parent, which is a large part of why the 21 is now the version most clinicians and most researchers reach for.

Every item describes a state and asks how much it applied over the past week, rated 0 (did not apply to me at all), 1 (applied to some degree, or some of the time), 2 (applied to a considerable degree, or a good part of the time), or 3 (applied very much, or most of the time). Twenty-one items take about three minutes. Nothing is reverse scored, which removes the single most common source of hand-scoring errors on instruments like the Perceived Stress Scale.

Depression covers dysphoria, hopelessness, devaluation of life, self-deprecation, loss of interest, anhedonia, and inertia. Notice what is absent: no sleep item, no appetite item, no concentration item, and no question about suicidal thinking. The scale was built to measure a mood construct rather than to reproduce a diagnostic criterion set, so somatic symptoms that overlap with medical conditions were deliberately left out. The practical consequence matters in the room. A DASS-21 depression score, at any level, is not a risk screen, and safety has to be asked about directly.

Anxiety covers autonomic arousal (dry mouth, difficulty breathing, trembling, awareness of heart action), skeletal muscle effects, situational anxiety, and the subjective experience of anxious affect. This scale reads the body. Compare it with the GAD-7, which is built around worry, restlessness, and irritability, and the difference in what the two instruments will detect becomes obvious: a client with heavy cognitive worry and little physiological arousal can score modestly on the DASS Anxiety scale while scoring high on the GAD-7. If you want that comparison in detail, see the guide to brief screening with the PHQ-9 and GAD-7. For clients who are unfamiliar with why their body reacts the way it does, Fight-or-Flight Response covers the physiology the Anxiety items are describing.

Stress covers difficulty relaxing, nervous arousal, being easily upset or agitated, irritability and over-reactivity, and impatience. Chronic non-specific arousal is the theme. Clients frequently recognize themselves in this scale faster than in the other two, because it describes the wound-up quality of a hard stretch without implying a disorder. Symptoms of Stress maps the same territory in worksheet form and pairs naturally with a high Stress score.

The three scales are interleaved through the questionnaire. Depression is items 3, 5, 10, 13, 16, 17, and 21. Anxiety is items 2, 4, 7, 9, 15, 19, and 20. Stress is items 1, 6, 8, 11, 12, 14, and 18. Sum the seven items belonging to each scale for a raw total between 0 and 21, and keep the three totals separate. There is no valid DASS-21 grand total for band interpretation, even though the 21 items summed together do function as a general distress index in research.

Now the step that gets missed constantly: multiply each raw scale total by 2. The doubled score, between 0 and 42, is what you compare against the published severity ranges. Nothing else in DASS scoring is complicated, and this one arithmetic move is where most misreadings start.

The reason is historical. The severity ranges were derived on the 42-item DASS, where each scale carried fourteen items and each scale total ran to 42. The DASS-21 halved the item count and halved the possible total with it. Doubling the short-form score puts it back on the original metric so the manual's cutoffs apply unchanged. The authors recommend exactly this, and every reputable scoring implementation does it silently in the background.

Skip the doubling and the errors run in one direction: everything reads far too mild. A raw depression sum of 12 is a severe score once doubled to 24, and it lands in the mild band if you forget. Anyone reviewing an outside report should check whether the number in front of them is a raw sum or a doubled score before saying anything about severity, particularly when the report shows a maximum of 21 rather than 42.

For a partially completed form, prorate rather than treating blanks as zeros. Compute the mean of the answered items on that scale, multiply by 7 to estimate the raw total, then double as usual. Two or more missing items on a seven-item scale makes the estimate shaky enough that re-administering is the better move.

Depression: 0 to 9 normal, 10 to 13 mild, 14 to 20 moderate, 21 to 27 severe, 28 and above extremely severe.

Anxiety: 0 to 7 normal, 8 to 9 mild, 10 to 14 moderate, 15 to 19 severe, 20 and above extremely severe.

Stress: 0 to 14 normal, 15 to 18 mild, 19 to 25 moderate, 26 to 33 severe, 34 and above extremely severe.

A worked example keeps this concrete. Say a client's raw sums come out at Depression 8, Anxiety 6, Stress 11. Doubling gives 16, 12, and 22. Read against the tables above, that profile is moderate depression, moderate anxiety, and moderate stress. Had you read the raw sums directly, you would have called it mild depression, mild anxiety, and normal stress, and the whole clinical picture would have shifted with it.

The three bands are set at different heights on purpose, which is why an anxiety score of 10 counts as moderate while a stress score of 10 sits comfortably inside normal. Community samples endorse tension and irritability freely and endorse trembling and breathing difficulty rarely, so the distributions the cutoffs were drawn from sit in different places.

This is the part that gets glossed over in most DASS explainers, and it changes how the number should be used. The DASS severity labels are norm-referenced. They were set by percentile position in a normative community sample: roughly, the normal label covers everything up to the 78th percentile, mild runs from the 78th to the 87th, moderate from the 87th to the 95th, severe from the 95th to the 98th, and extremely severe covers the top 2 percent.

So when a DASS-21 result reads moderate anxiety, the underlying claim is positional. This person's anxiety score sits somewhere between the 87th and 95th percentile of a general community sample. That is a statement about a distribution, and it says nothing directly about whether a diagnosis is present, whether treatment is indicated, or whether a threshold established against a structured interview has been crossed.

PHQ-9 severity works from the opposite direction. Its bands are anchored to symptom counts drawn from the DSM criteria for major depressive disorder, and the score of 10 that opens its moderate band was validated against diagnostic interviews, where Kroenke and colleagues (2001) reported sensitivity and specificity both around 88 percent for major depression. A PHQ-9 of 10 is a criterion-anchored screening threshold. A DASS-21 depression score of 14 is a percentile marker. Two words that sound identical on a report are answering different questions.

One consequence catches clinicians off guard. The DASS manual notes that a mild label means the person scores above the population average while still, in all likelihood, falling below the typical severity of someone who seeks treatment. A mild label therefore describes someone sitting a little above average on a dimensional measure of distress, which is a long way from a mild disorder.

The second consequence is about cross-instrument comparison. A client who scores moderate on both the DASS-21 depression scale and the PHQ-9 has produced agreement that looks stronger than it is, since one number is a percentile and the other is a symptom count. Read each against its own scale and its own reference frame, and resist the urge to average them into a single impression.

The third consequence is about the norms themselves. Percentile labels inherit whatever sample they came from. The original cutoffs come from Australian community data, and Henry and Crawford (2005) later published UK norms on a large non-clinical sample. Where a client belongs to a population with a meaningfully different distribution, the label carries a little less weight than the raw number does.

None of this makes the DASS a weaker instrument. Dimensional measurement is what it was built for, and it tracks change over a course of treatment better than most criterion-anchored screens do, precisely because it is not trying to sort people into cases and non-cases. The severity labels are a convenient shorthand layered on top of a continuous score, and the continuous score is the part that carries the clinical information.

Henry and Crawford (2005) administered the DASS-21 to 1,794 non-clinical UK adults (815 men and 979 women, ages 18 to 91) and reported internal consistency of .88 for Depression, .82 for Anxiety, .90 for Stress, and .93 for the full 21 items. Their factor analysis supported a structure with a general distress factor sitting above the three specific factors, which is the empirical version of the clinical observation that these three constructs travel together while still being separable.

The Anxiety scale carries the lowest alpha of the three, and the reason is visible in the item content. Seven items are being asked to cover autonomic arousal, muscle tension, situational fear, and subjective anxious affect at once. That breadth costs some internal consistency and buys coverage of presentations a narrower scale would miss.

Antony and colleagues (1998) tested the instrument in clinical groups (panic disorder, social phobia, specific phobia, obsessive-compulsive disorder, and major depression) alongside a community sample, reporting internal consistency above .85 on all three scales and a three-component solution that accounted for 67 percent of item-level variance. Their finding that the 21-item version factored more cleanly than the 42-item original is the practical case for using the short form.

Convergent and discriminant evidence comes from the 1995 Behaviour Research and Therapy paper, where the DASS scales were compared against the Beck Depression Inventory and the Beck Anxiety Inventory. The DASS Anxiety scale correlated strongly with the BAI, the DASS Depression scale with the BDI, and the Stress scale captured variance neither Beck measure isolated.

Test-retest performance over roughly two weeks sits in a range typical for state measures with a past-week window, which is what you want from an instrument used for repeated administration. The scores are meant to move.

0 to 9 (Normal): the large majority of a community sample scores in this range, since the band extends to roughly the 78th percentile. A normal score does not rule out depression in someone who minimizes on self-report, and it does not speak to risk at all, because the scale contains no item about suicidal thinking. Where the interview and the number disagree, follow the interview.

10 to 13 (Mild): above the population average and, per the manual's own framing, still probably below the severity of a typical help-seeker. Psychoeducation, behavioral activation, attention to sleep and load, and a repeat administration in a few weeks are the usual responses. A mild score that stays put across three or four administrations means more than a single reading a few points higher.

14 to 20 (Moderate): the 87th to 95th percentile range. A DASS-21 depression score of 16 after doubling falls in the moderate range, which corresponds to a raw sum of 8 before doubling. Scores here usually reflect endorsement of hopelessness or loss of interest at the higher response options, and they warrant a diagnostic conversation rather than a diagnosis. Pair the number with a criterion-anchored measure if the question on the table is whether a depressive episode is present.

21 to 27 (Severe): the 95th to 98th percentile. Dysphoria and anhedonia are being endorsed near the top of the response scale across most of the seven items. Active treatment, direct assessment of functioning, and a shorter follow-up interval are standard. Ask about suicidal ideation explicitly, since nothing in the score can substitute for that question.

28 and above (Extremely Severe): the top 2 percent of the normative distribution. Assume substantial impairment until the interview says otherwise, and treat safety assessment as the first item of business rather than a closing formality. If you or someone you know is in crisis, contact the 988 Suicide and Crisis Lifeline by calling or texting 988.

0 to 7 (Normal): community samples endorse the autonomic items sparingly, so this band is narrower than its Depression and Stress counterparts even though it covers the same 78 percent of the distribution. A raw sum of 3 doubles to 6 and stays inside normal; a raw sum of 4 doubles to 8 and does not.

8 to 9 (Mild): two points wide, which makes it the most fragile band on the instrument. A single item moving from a 1 to a 2 can carry someone out of normal and into mild, and a second such shift can carry them into moderate. Read scores near these boundaries as approximate, and give more weight to which items are elevated than to which label the total landed in.

10 to 14 (Moderate): a DASS-21 anxiety score of 12 after doubling falls in the moderate range. Expect the client to be reporting real physiological arousal: heart awareness, breathing difficulty, trembling, or a sense of being close to panic. This is the band where the specific items earn their keep, because panic-type arousal and anticipatory situational anxiety call for different first interventions. A Countering Anxious Thoughts log is a reasonable between-session assignment while the picture is still being clarified.

15 to 19 (Severe): near-ceiling responses across most items. Screen for panic disorder specifically, ask about avoidance, and ask about alcohol and substance use as self-management, since severe autonomic arousal frequently comes with an attempt to switch it off. Comorbid depression is common enough at this level to be worth measuring rather than assuming.

20 and above (Extremely Severe): the top 2 percent. Consider medical contributors before settling on a purely psychological formulation, since hyperthyroidism, stimulant use, withdrawal states, and cardiac and respiratory conditions all push these particular items up. Coping Skills: Anxiety gives a client something concrete to work with while the assessment is underway.

0 to 14 (Normal): the widest normal band of the three, and for a good reason. Ordinary people report tension, impatience, and difficulty winding down at a much higher base rate than they report trembling or hopelessness, so the distribution these cutoffs were drawn from sits higher. A doubled stress score of 12 is unremarkable in a way that a doubled anxiety score of 12 is not.

15 to 18 (Mild): a modest elevation above the community average. Where circumstances explain it, load reduction and recovery time are usually the whole intervention. Stress Exploration is a useful way to get the specific demands named rather than left as a general sense of too much.

19 to 25 (Moderate): a doubled stress score of 19 falls at the bottom of the moderate band, and 25 sits at its top. This range describes someone reactive, wound up, and having trouble getting back to baseline between demands. Scores here often precede a rise on the Depression scale by weeks or months, so a moderate stress result in an otherwise normal profile is information about trajectory.

26 to 33 (Severe): sustained arousal at this level tends to be visible in sleep, in relationships, and in work performance. Treat it as a target rather than as context for the other two scales, and check whether the Depression and Anxiety scales are following it upward on repeat administration.

34 and above (Extremely Severe): the top 2 percent. Where this score appears alongside normal Depression and Anxiety scores, the presentation is usually load and reactivity without a mood syndrome, and preventive work is the right frame. Where all three are elevated, the general distress factor is doing most of the work and the shape of the profile tells you less than its overall height.

The profile is the point. A single elevated scale is easy to interpret; the value of the DASS-21 shows up in the pattern across all three, which a single-construct measure cannot produce.

High Stress with normal Depression and Anxiety is the most common clinically interesting pattern, and it usually describes a person under real load whose regulation is still holding. Load reduction, recovery, and shoring up coping resources are the work. Reading this profile as a disorder in early form misdescribes what is happening.

High Anxiety with a normal Stress score is unusual and worth a second look. Because the Stress scale captures the general arousal component, a high Anxiety score without it points toward discrete, situationally bound arousal: panic episodes, a specific phobia, or performance anxiety that leaves the rest of the week untouched. Decatastrophizing fits this presentation well once the feared outcome has been named.

High Depression with high Stress and only modest Anxiety shows up frequently in the burnout-toward-depression trajectory, where months of unmanaged load have started to erode interest and hope while the body has never been especially aroused. The Stress scale usually moves first in both directions, which makes it the early indicator on the way in and the early sign of response on the way out.

Because the items ask about the past week, weekly re-administration measures a clean non-overlapping window, and session-by-session use is defensible in a way it is not for two-week instruments like the PHQ-9 or month-long ones like the PSS-10 and K10, which are covered in the guide to measuring stress and distress. Keep the interval constant, because totals gathered over different windows are not really comparable.

Read movement conservatively. A shift that crosses a band boundary and holds across two administrations means considerably more than a few points of wobble inside a band, and on the Anxiety scale in particular, where the bands are only a few points wide, a single item's movement can cross a boundary without anything clinically meaningful having changed.

The DASS questionnaire is public domain, and its authors at the University of New South Wales state plainly that permission is not needed to use it. What they also state is a condition on automated delivery: computed scores should not be shown to respondents, and automated interpretation should not be attempted, on a website open to the public. Where a site is restricted to a defined group such as enrolled patients or research participants, and results go to the clinician or researcher rather than back to the respondent, automated administration is fine by their account.

Therapy Resource follows that condition. The researchers who build and validate an instrument set the terms of its use, and honoring those terms is basic professional ethics. So this article explains how DASS-21 scoring and interpretation work, it does not host the scale or score it for you, and clinicians who want the questionnaire, the scoring template, and the manual can download them directly from the official DASS site at UNSW at no cost.

Readers who came here wanting a free instrument they can complete and have scored immediately are well served elsewhere on this site. The free assessments include the PHQ-9, the GAD-7, the K10, and the PSS-10, all scored and banded on the spot with no account required. Those four cover depression, anxiety, general distress, and perceived stress, which is most of what the DASS-21 covers, using instruments whose licensing permits exactly this kind of delivery.

What does my DASS-21 score mean? You have three separate scores, one for each scale, and each one gets doubled before it is compared with its own severity table. The label attached to each doubled score describes where you sit relative to a community sample: normal covers roughly the bottom 78 percent, and each label above it marks a higher slice of that distribution. Read the three numbers side by side rather than adding them, and take any label as a starting point for a conversation with a clinician.

Do I multiply DASS-21 scores by 2? Yes. Sum the seven items on a scale, then double that sum before looking at any severity range. The published cutoffs were established on the 42-item DASS, where each scale ran to 42 points rather than 21, and doubling the short-form score restores that metric. Comparing an undoubled DASS-21 total against those cutoffs makes almost every result look one or two bands milder than it is.

Is the DASS-21 a diagnosis? No. It measures severity along three dimensions over the past week, and it was built without reference to any diagnostic criterion set, which is why it contains no sleep item, no appetite item, and no suicide item. Diagnosis requires a clinical interview covering symptom count, duration, functional impairment, history, and the medical and substance-related explanations that produce similar answers. The DASS-21 tells you how much distress a person is carrying and in what proportions. Naming the condition is a separate task.

Is a DASS-21 depression score of 16 bad? A 16 after doubling sits in the moderate band, between the 87th and 95th percentile of a community sample. It describes distress well above average without reaching the top of the scale, and it is a reasonable prompt for a diagnostic conversation. The word moderate here is a percentile position rather than a clinical verdict.

Why are the anxiety cutoffs so much lower than the depression and stress cutoffs? Because the bands come from percentiles rather than from a common metric. People in community samples rarely endorse trembling, breathing difficulty, and heart awareness, so the anxiety distribution is compressed toward zero and the percentile boundaries land lower. A doubled anxiety score of 10 and a doubled stress score of 22 both sit in the moderate band, and both mean roughly the same thing about position in the distribution.

Can a DASS-21 score be compared with a DASS-42 score? Yes, once the DASS-21 scale totals have been doubled. That is the entire purpose of the doubling step. Treat the comparison as close rather than exact, since the short form samples seven of the original fourteen items per scale.

How often should the DASS-21 be re-administered? The past-week window supports weekly administration during active treatment, and monthly is reasonable during maintenance. Keep whatever interval you choose constant, and watch band boundaries with some skepticism on the Anxiety scale, where two points separate several of them.

Lovibond, S. H., and Lovibond, P. F. (1995). Manual for the Depression Anxiety Stress Scales (2nd ed.). Sydney: Psychology Foundation of Australia.

Lovibond, P. F., and Lovibond, S. H. (1995). The structure of negative emotional states: Comparison of the Depression Anxiety Stress Scales (DASS) with the Beck Depression and Anxiety Inventories. Behaviour Research and Therapy, 33(3), 335-343.

Henry, J. D., and Crawford, J. R. (2005). The short-form version of the Depression Anxiety Stress Scales (DASS-21): Construct validity and normative data in a large non-clinical sample. British Journal of Clinical Psychology, 44(2), 227-239.

Antony, M. M., Bieling, P. J., Cox, B. J., Enns, M. W., and Swinson, R. P. (1998). Psychometric properties of the 42-item and 21-item versions of the Depression Anxiety Stress Scales in clinical groups and a community sample. Psychological Assessment, 10(2), 176-181.

Kroenke, K., Spitzer, R. L., and Williams, J. B. W. (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16(9), 606-613.