Rating scales
Translate a total between the Hamilton Depression Rating Scale and the Montgomery-Åsberg scale using the published equipercentile linking — with the severity band and remission status on both scales, and the percent-change figure that needs no conversion at all.
Free · No sign-up · Runs entirely in your browser
Severity band
Severe
Reference band, not a diagnosis
Remission (≤7)
No
Frank et al., 1991
MADRS total score
31
31.2 unrounded · 0–60 · Montgomery-Åsberg Depression Rating ScalePopulation-level approximation — not a MADRS rating of this person.Severity band
Moderate
Reference band, not a diagnosis
Remission (≤10)
No
Hawley et al., 2002
Add the pre-treatment total to convert an improvement rather than a level — the paper tabulates the two separately, and they do not use the same ratio.
These are rank-order equivalents, not interchangeable scores. Equipercentile linking matches distributions: it identifies the MADRS total sitting at the same percentile as your HAM-D total in the 4,388-patient trial population Leucht and colleagues studied. It does not tell you the score this patient would have received had the other scale been administered, and the two instruments do not measure quite the same thing — the HAM-D devotes three items to insomnia and several to somatic symptoms, where the MADRS has one sleep item and concentrates on core mood and cognition. Report the scale you actually used, and present a converted value as an approximation with its source.
Plain text: both totals with their bands and remission status, the improvement in both currencies, the percent change, and the caveat.
Conversions follow the equipercentile tables of Leucht, Fennema, Engel, Kaspers-Janssen & Szegedi (2018), estimated from 4,388 patients in 31 mirtazapine trials in major depressive disorder. Values between published anchors are interpolated; values outside them are not estimated at all. Severity bands come from each scale’s own convention (Zimmerman et al., 2013 for the HAM-D; Snaith et al., 1986 for the MADRS). This is an aid for education and research support, not a diagnosis.
Reference · for the curious
Two clinician-rated depression scales dominate the literature, and they are not the same scale. The Hamilton Depression Rating Scale has been the field's default since 1960; the Montgomery-Åsberg scale was designed in 1979 specifically to be more sensitive to change during treatment. Meta-analyses, treatment guidelines and any attempt to compare two trials eventually run into the problem this tool addresses: a total on one has to be read against a total on the other.
Leucht and colleagues (2018) took 4,388 patients from 31 mirtazapine trials in major depressive disorder, all rated on both instruments at the same visits — baseline and weeks 1, 2 and 4 — and matched the two score distributions. The result is a correspondence table: for any HAM-D total, the MADRS total that sits at the same percentile of the same population.
The published anchors are HAM-D 10, 20, 30 and 40 corresponding to MADRS 13, 26, 39 and 52–53. That is close to a constant ratio of about 1.3, and you will see 1.3 used as a rule of thumb, but the tables are the citable object and a multiplier is a simplification of them.
The critical thing this method does not do is tell you what an individual patient would have scored on the other instrument. It is a statement about distributions in a population, not about people. Two patients with an identical HAM-D of 22 can sit several MADRS points apart, because the two scales weight symptoms differently: the HAM-D spends three of its seventeen items on insomnia and several more on somatic complaints, while the MADRS has a single sleep item and concentrates on mood, interest and cognition. Our MADRS versus HAM-D comparison works through those differences item by item, and depression rating scales compared sets both against the PHQ-9 and content-analytic approaches.
The same paper reports a finding that is more useful than the conversion table itself: a percentage improvement from baseline on the HAM-D is approximately the same percentage improvement on the MADRS. Percent change transfers between the two scales without any table at all, and therefore without interpolation error or a range restriction.
So if your question is whether two trials produced comparable improvement, compare percent reductions and cite that finding. Reach for the total-score table only when you genuinely need a level — an entry criterion, a severity band, a remission threshold.
Absolute improvements are a third case, and they need their own table. A 10-point HAM-D improvement corresponds to about 12 MADRS points, a 20-point improvement to 26, and a 25-point improvement to 34. Those ratios (1.20, 1.30, 1.36) are not the constant the total-score anchors imply, which is precisely why the paper tabulates change separately. The calculator above uses the change table for improvements and the total table for levels, and will not silently substitute one for the other.
HAM-D remission is conventionally a total of 7 or below; MADRS remission is conventionally 10 or below. Those two cut-points were derived independently, by different investigators, in different literatures. Run the HAM-D threshold through the linking and it converts to a MADRS of about 9 — roughly a point stricter than the MADRS's own convention.
Two separate lines of work landing that close together is mild but real evidence that both thresholds identify a similar clinical state. It also means remission rates from HAM-D and MADRS trials are more comparable than response rates, which depend on baseline severity.
The published anchors span HAM-D totals from 0 to 40. The HAM-D-17 itself runs to 52, but the trials the linking came from were of major depressive disorder, where totals above 40 are rare, so the upper range is unanchored. The calculator returns nothing there rather than extending the line, because a number produced past the last anchor would carry the authority of a citation that does not cover it.
The same applies in reverse: a MADRS above about 53 has no HAM-D equivalent in these tables. And a score above a scale's own maximum is treated as a data-entry error rather than converted at all.
Conversion is a way of reading other people's data. For your own study, rating both instruments from the same interview is strictly better than rating one and converting, and it is less work than it sounds: the two scales overlap heavily in the symptoms they ask about, and a single well-structured interview supports both. What it requires is that each item rating be traceable to something specific, rather than assigned from a global impression formed at the end.
That is the case for auditable scale scoring — tying every item to the passage that justifies it — and stacking several instruments on one transcript covers doing it for more than one scale at a time. Score either instrument item by item with the HAM-D calculator or the MADRS calculator, and if two raters are involved, check agreement with the inter-rater reliability calculator. Both instruments are described in the Tagaroo Scale Library.
Both scales are freely reproducible with citation: the HAM-D was published by Max Hamilton in 1960 and the MADRS by Stuart Montgomery and Marie Åsberg in 1979. The equipercentile tables are the intellectual property of their authors and are cited here rather than reproduced — this tool encodes the anchor values the paper states in its results, and interpolates between them.
This is a conversion aid for education and research support. It does not diagnose, and a converted score is an approximation from population data rather than a measurement of the person in front of you. Everything is computed in your browser; no scores are transmitted or stored.
Using equipercentile linking. Leucht and colleagues (2018) took 4,388 patients from 31 mirtazapine trials in major depressive disorder, all rated on both scales simultaneously, and matched the two distributions. HAM-D totals of 10, 20, 30 and 40 corresponded to MADRS totals of roughly 13, 26, 39 and 52–53 — close to a constant ratio of about 1.3, though the paper's own tables are what should be quoted rather than a multiplier. This calculator interpolates between those published anchors and declines to produce a value outside them.
No, and this is the most important limitation to state. Equipercentile linking gives rank-order correspondence between distributions: it identifies the MADRS total sitting at the same percentile as a given HAM-D total, in the population studied. It does not tell you the score an individual patient would have received had the other instrument been administered. The two scales also weight symptoms differently — the HAM-D has three insomnia items and several somatic ones, the MADRS has one sleep item and concentrates on core mood and cognition — so two patients with the same HAM-D total can differ on the MADRS. Report the scale you actually administered, and present any converted figure as an approximation with its citation.
Percent change, whenever it will answer your question. The same study found that a percentage improvement from baseline on the HAM-D is approximately the same percentage improvement on the MADRS — so percent change transfers between the scales without any conversion table, and without interpolation error. That makes it the right currency for comparing trials that used different instruments. Absolute improvements need the paper's separate change table: a 10-point HAM-D improvement corresponds to about 12 MADRS points, not the 13 the total-score ratio would suggest, because level and change do not scale identically.
Almost exactly, which is a useful cross-check. HAM-D remission is conventionally a total of 7 or below and MADRS remission a total of 10 or below; those cut-points were derived separately, in different literatures. Running the HAM-D threshold of 7 through the equipercentile linking gives a MADRS equivalent of about 9 — roughly one point stricter than the MADRS's own conventional cut-point. Two independent lines of work landing that close together is mild evidence that both thresholds are picking out a similar clinical state.
Because the linking does not cover them. The published anchors span HAM-D totals of 0 to 40; the trials it was estimated from were of major depressive disorder, where HAM-D-17 totals above 40 are rare, so there is no reliable correspondence up there even though the scale itself runs to 52. Extending the line past the last anchor would produce a number that looks as though the paper supported it. The tool shows nothing and explains the range instead. Scores above a scale's own maximum are reported as data errors rather than converted.
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 scores the HAM-D and the MADRS against the same transcript, with every item rating tied to the passage that supports it — so if you need both instruments, you can rate both from one interview instead of converting between them.
HAM-D ↔ MADRS converter · tagaroo.ai/materials/hamd-madrs-converter · figures verified against primary sources 2026-07-29. Educational scoring aid, not a diagnosis, and not reviewed by a licensed clinician.