Rating scales
Score the Hamilton Anxiety Rating Scale: fourteen clinician-rated items on 0–4, a 0–56 total, the psychic and somatic subscales reported separately — and a toggle between the two severity conventions that disagree about what your total means.
Free · No sign-up · Runs entirely in your browser
Rate each symptom cluster for the past week. 0 = Not present, 1 = Mild, 2 = Moderate, 3 = Severe, 4 = Very severe / grossly disabling.
Worry, anticipation of the worst, fearful anticipation, irritability
Inability to relax, startling, trembling, restlessness, easily moved to tears, fatigability
Of the dark, strangers, being left alone, animals, traffic, crowds
Difficulty falling asleep, broken sleep, unrefreshing sleep, dreams, nightmares, night terrors
Difficulty concentrating, poor memory
Loss of interest, lack of pleasure in hobbies, low mood, early waking, diurnal swing
Aches and pains, twitching, stiffness, myoclonic jerks, teeth-grinding, unsteady voice, increased muscle tone
Tinnitus, blurred vision, hot and cold flushes, feelings of weakness, pricking sensations
Tachycardia, palpitations, chest pain, throbbing of vessels, faintness, missed beats
Chest pressure or constriction, choking sensations, sighing, breathlessness
Difficulty swallowing, wind, abdominal pain or burning, fullness, nausea, loose bowels, constipation, weight loss
Urinary frequency or urgency, menstrual disturbance, loss of libido, sexual dysfunction
Dry mouth, flushing, pallor, sweating, giddiness, tension headache, raised hair
Observed only: fidgeting, restlessness, hand tremor, furrowed brow, strained face, sighing, rapid breathing, pallor, swallowing
Total score
0 / 56
Sum of all fourteen items
Severity band
No or minimal anxiety
Matza et al., 2010 · reference band, not a diagnosis
Psychic anxiety
0 / 28
Items 1–6 and 14
Somatic anxiety
0 / 28
Items 7–13 · 0% of total
Enter the pre-treatment total to see change from baseline and the 50% response criterion.
Response follows the ≥50% reduction criterion (Frank et al., 1991). There is no remission reading here on purpose: unlike the HAM-D (≤7) and MADRS (≤10), the HAM-A has no single agreed remission cut-point — trials commonly use ≤7, but not consistently, so report the definition your protocol specifies. A total at or above 8 is the lowest that is more than minimal on the empirical bands.
Plain text: every item with its code and anchor, the total, both subtotals, the convention you chose, and — when a baseline is entered — the change from it.
Items follow Hamilton (1959); the psychic and somatic groupings follow Maier et al. (1988); the default bands follow Matza et al. (2010). This is a scoring aid for education and research support, not a diagnosis — the HAM-A quantifies the severity of anxiety and does not establish which anxiety disorder, if any, is present. Rate the items above, or load the example, to see the score update.
Reference · for the curious
The Hamilton Anxiety Rating Scale is the oldest anxiety severity measure still in routine use. Max Hamilton published it in 1959, four years before the depression scale that carries his name, and it remains the conventional severity endpoint in anxiety treatment trials. Fourteen items, each rated 0 to 4 by a clinician, for a total of 0 to 56.
Thirteen items cover symptom clusters the person reports — anxious mood, tension, fears, insomnia, cognitive difficulty, depressed mood, and seven groups of bodily symptom. The fourteenth, behaviour at interview, is rated purely from what you observe: fidgeting, tremor, a strained face, sighing respiration. There is no self-report version. The HAM-A was designed as a clinician rating and its reliability depends on a trained interviewer, which is the first thing to report about any HAM-A dataset.
This is the scale's most practical trap. Most printed copies of the form carry the older convention: under 17 is mild, 18–24 mild to moderate, 25–30 moderate to severe. Read it carefully and two gaps appear — it labels nothing at all above 30, and writing the first edge as “under 17” leaves 17 itself unassigned.
Matza and colleagues (2010) derived empirical bands against clinician global impressions: 7 or below no or minimal anxiety, 8–14 mild, 15–23 moderate, 24 or above severe. These cover the whole range without gaps, which is why the calculator above defaults to them.
The two disagree over most of the scale. A total of 18 is “moderate” empirically and “mild to moderate” on the form; a total of 25 is “severe” empirically and “moderate to severe” on the form. Neither is wrong, but they are not the same claim, and HAM-A severity proportions are not comparable across studies that used different bands. The toggle in the tool shows both readings of the same total, and tells you when they part company.
Factor analyses of the HAM-A, most influentially Maier and colleagues (1988), split it in two. Psychic anxiety is items 1–6 and 14; somatic anxiety is items 7–13. That means exactly half the scale asks about bodily sensations: muscular aches, tinnitus and blurred vision, palpitations, breathlessness, nausea, urinary frequency, sweating and flushing.
The consequence is the HAM-A's best-known confound. A person with a cardiac condition, a thyroid disorder, or the side effects of a medication can accumulate a substantial somatic subtotal from symptoms their physical illness already explains, and land in an anxious band without much psychic anxiety at all. The same total means something different depending on where it came from, so the calculator reports the two subtotals alongside the sum and flags the cases where somatic items are carrying it.
This is also why the HAM-A moves in antidepressant trials for reasons that have little to do with anxiety, and why the psychic subscale is often the more sensitive outcome. If you are reporting change on the HAM-A, report the subscales too.
They do different jobs. The GAD-7 is seven self-report items, takes about a minute, and screens: 10 or above means an anxiety disorder is probable. The HAM-A is fourteen clinician-rated items, takes fifteen to twenty minutes, and quantifies severity once you already know anxiety is present. Use the GAD-7 to decide whether to look harder; use the HAM-A to measure and to track change.
They correlate strongly, but a score on one is not a score on the other, and converting between them is a rank-order approximation rather than a translation. Our comparison of GAD-7 versus HAM-A works through when the extra cost of a clinician-rated scale earns its keep, and clinician-rated versus self-report scales covers the general trade-off between who rates and what you can conclude.
Eleven of the fourteen items can be grounded in something the subject says, which is what makes the HAM-A workable as a transcript instrument. Three cannot: behaviour at interview is observational by definition, and the sensory and genitourinary clusters are frequently unmentioned in a clinical interview unless asked about directly. Decide before coding whether an unmentioned cluster is a zero or a missing value, and record that decision — it is the single largest source of disagreement between raters on this scale.
Write down what counts as evidence for each item first. Our guide to the HAM-A's items, subscales and cutoffs covers the anchors item by item, and turning a rating scale into a codebook covers making the mapping explicit enough that two people apply it the same way. If two raters are scoring the same transcripts, compute agreement with the inter-rater reliability calculator — and compute it on the subscales as well as the total, because a respectable total agreement can hide raters who disagree systematically about which items are carrying it.
Maier and colleagues (1988) reported good inter-rater reliability for the total score and acceptable internal consistency, along with the finding that the psychic subscale was more sensitive to treatment change than the total. Later work has been more critical of the instrument's dimensionality: the items were never designed as a factor-analytic structure, the somatic items overlap with medical symptoms, and several clusters bundle symptoms that do not reliably co-occur. Treat the total as a severity index with known confounds rather than a measurement of one thing, and report your rater training.
The HAM-A is in the public domain — Hamilton published it in the British Journal of Medical Psychology in 1959 — and is freely reproducible with citation. It is hosted in the Tagaroo Scale Library for that reason. Commercial rater-training packages and licensed electronic implementations exist, but they do not restrict the instrument itself.
This is a scoring aid for education and research support. It does not diagnose, it does not distinguish one anxiety disorder from another, and it does not replace clinical assessment. Everything is computed in your browser; no ratings are transmitted or stored.
Fourteen items are each rated 0 (not present) to 4 (very severe, grossly disabling) by a clinician, giving a total from 0 to 56. Thirteen items cover symptom clusters the person reports — anxious mood, tension, fears, insomnia, cognitive difficulty, depressed mood, and seven groups of bodily symptom — and the fourteenth, behaviour at interview, is rated purely from observation. There is no self-report version of the HAM-A: it was designed as a clinician rating and its reliability depends on a trained interviewer.
On the empirical bands derived by Matza and colleagues (2010), a HAM-A total of 7 or below is no or minimal anxiety, 8–14 is mild, 15–23 is moderate, and 24 or above is severe. Those are the bands this calculator uses by default, because they cover the full 0–56 range. Be aware that a second convention is still in circulation and disagrees with them: the one printed on most copies of the form reads under 17 as mild, 18–24 as mild to moderate, and 25–30 as moderate to severe — and labels nothing at all above 30. A total of 18 is therefore "moderate" on the empirical bands and "mild to moderate" on the printed form. Always state which convention you applied; HAM-A severity rates are not comparable across studies that use different bands.
Factor analyses, most influentially Maier and colleagues (1988), split the scale in two. Psychic anxiety covers items 1–6 and 14: anxious mood, tension, fears, insomnia, cognitive difficulty, depressed mood and observed behaviour. Somatic anxiety covers items 7–13: muscular, sensory, cardiovascular, respiratory, gastrointestinal, genitourinary and autonomic symptoms. The split matters because half the scale asks about bodily sensations, so a physically unwell person can reach an anxious-looking total on symptoms their illness already explains. Reporting the two subtotals alongside the total makes the composition of a score visible instead of leaving it inside the sum.
They answer different questions. The GAD-7 is seven self-report items, takes about a minute, and is built to screen: a total of 10 or above indicates that an anxiety disorder is probable. The HAM-A is fourteen clinician-rated items, takes fifteen to twenty minutes, gives more resolution and an explicit psychic/somatic split, and is the conventional severity endpoint in anxiety trials. Use the GAD-7 to decide whether to look harder, and the HAM-A to quantify severity or track change once you already know anxiety is present. They correlate strongly but are not interchangeable, so a score on one cannot be reported as a score on the other.
Yes, and it is the HAM-A's best-known confound. Seven of the fourteen items ask about bodily symptoms — muscular, sensory, cardiovascular, respiratory, gastrointestinal, genitourinary and autonomic — so half the available points can come from sensations a physical illness or a medication side effect already explains. A person with a cardiac condition, a thyroid disorder or a drug causing dry mouth and sweating can accumulate a substantial somatic subtotal and land in an anxious band with little psychic anxiety at all. This is also why the HAM-A moves in drug trials for reasons unrelated to anxiety, and why the psychic subscale is often the more sensitive outcome. Report the psychic and somatic subtotals alongside the total so the composition of a score is visible, not just its size.
Yes. Max Hamilton published the scale in 1959 in the British Journal of Medical Psychology and it is in the public domain, freely reproducible with citation. Note that some commercial rater-training packages and licensed electronic implementations of the HAM-A do exist; the instrument itself is not restricted by them.
Written by Enrique Gutiérrez, PhD (Computer Science) — founder of Tagaroo and Associate Professor of Computer Science, working on inter-rater reliability, measurement and annotation methodology (ORCID).
How this page is checked. Scoring rules, item ranges and thresholds are transcribed from the instrument's cited primary sources and covered by automated tests that reproduce each paper's own worked examples. This page has not been reviewed by a licensed clinician, and it is a scoring aid for education and research support rather than a clinical decision tool.
Last verified: 29 July 2026. Formulas, thresholds and cited figures on this page were checked against their original sources on that date. Every calculation runs in your browser; nothing you enter is transmitted or stored.
Tagaroo lets you score the HAM-A against a transcript, with every item rating tied to the passage that supports it — so a psychic/somatic split you can defend, item by item, instead of a remembered global impression.
HAM-A calculator · tagaroo.ai/materials/ham-a-calculator · figures verified against primary sources 2026-07-29. Educational scoring aid, not a diagnosis, and not reviewed by a licensed clinician.