ICH Score Calculator

Score the five items of the intracerebral haemorrhage score, 0 to 6, and read the 30-day mortality each stratum carried in the single-centre cohort it was derived in. A derivation cohort is not a prognosis for this patient.

Last updated 8 Oct 2026

ICH score

Five items, 0 to 6

Glasgow Coma Scale on presentationThe GCS at presentation, before intubation and sedation. This is the only two-point item and the one most often taken from the wrong moment: a post-intubation score is not the presenting score, and is not a valid GCS at all once a component is untestable. See the Glasgow Coma Scale calculator for the NT rule.

Age 80 years or overA step at 80, not a gradient: nothing at 79 and one point at 80. A crude treatment of age and a recognised limitation.

Haematoma volume 30 cm³ or moreVolume on the presenting CT, measured in the derivation paper by ABC/2. That matters: ABC/2 overestimates by about 20 per cent on average against planimetry, so a planimetric 26 mL can be an ABC/2 31 mL and cross this threshold. Use the ABC/2 volume calculator and read its error section before scoring a borderline clot.

Intraventricular haemorrhageAny blood in the ventricular system on the presenting CT. The score makes no distinction between a trace of layering in an occipital horn and a cast filling all four ventricles, which is what the Graeb and Hijdra scales were written to capture.

Infratentorial originOrigin, not extension. A thalamic haemorrhage that has tracked into the midbrain is supratentorial in origin and scores 0. The item exists because a small volume in the posterior fossa behaves like a much larger one above the tentorium.

3pointsExample

GCS 9 at presentation, age 62, haematoma volume 38 cm³ with intraventricular extension, left putaminal origin

Scoring

ICH score = GCS (0–2) + age 80 or over (0–1) + volume 30 cm³ or more (0–1) + intraventricular haemorrhage (0–1) + infratentorial origin (0–1)
Range 0 to 6

the GCS item: the only two-point item. GCS 13–15 scores 0, 5–12 scores 1, 3–4 scores 2. The strata are not the usual mild / moderate / severe bands: a GCS of 5 and a GCS of 12 score the same point here
the derivation: a retrospective review of all patients presenting with acute intracerebral haemorrhage to the University of California, San Francisco during 1997 and 1998. One centre, two years, before any of the modern blood-pressure, haemostatic or minimally invasive evacuation trials
what the strata predicted: 30-day mortality of 0, 13, 26, 72, 97 and 100 per cent at scores of 0 to 5. The published anchors are exact and small: “all 26 patients with an ICH Score of 0 survived, and all 6 patients with an ICH Score of 5 died”. A denominator of 6 at the top is worth holding in mind before quoting 100 per cent, and a score of 6 did not occur at all
what it is not for: limiting treatment. It was designed to standardise severity stratification at presentation so that series and trials could be compared. The 2022 AHA/ASA guideline states that baseline severity scales “should not be used as the sole basis for limiting life-sustaining treatments”

Worked example

GCS 9 at presentation, age 62, haematoma volume 38 cm³ with intraventricular extension, left putaminal origin
GCS 9 falls in the 5 to 12 stratum = 1
Age 62 = 0 · volume 38 cm³, at or above 30 = 1 · intraventricular haemorrhage = 1 · supratentorial = 0
1 + 0 + 1 + 1 + 0 = 3 points
Thirty-day mortality at 3 in the derivation cohort was 72 per cent; in Malinova’s treated cohort it was 1 of 82, or 1.3 per cent. Both describe their own cohorts and neither is this patient’s probability
Watch how little it takes to move. A presenting GCS of 4 rather than 9 gives 4 and a derivation figure of 97 per cent. A planimetric volume of 29 cm³ rather than an ABC/2 38 gives 2 and 26 per cent. One item is the difference between 26 and 97 per cent

The five items

Every item is read off the presenting assessment and the presenting CT. Two of the five are measured rather than observed, and both are soft: the GCS is often recorded after intubation, and the 30 cm³ threshold sits well inside the measurement error of the ABC/2 method the derivation used.
Item	0 points	1 point	2 points
Glasgow Coma Scale	13 to 15	5 to 12	3 to 4
Age	Under 80 years	80 years or over	—
Haematoma volume	Under 30 cm³	30 cm³ or more	—
Intraventricular haemorrhage	Absent	Present	—
Origin	Supratentorial	Infratentorial	—

Thirty-day mortality by score, and who reports what

Two things this table is for: the disagreement at a score of 4, where most reproductions give 97 per cent and at least one gives 94 from the same derivation; and the size of the gap between the columns. Malinova’s cohort was selected and treated, which is why the ICH score should be read as a severity ranking rather than a mortality prediction.
Score	Derivation cohort (UCSF, 1997 to 1998)	Treated cohort, n = 233 (Malinova 2020)
0	0% — all 26 patients survived	not reported
1	13%	0% (0 of 13)
2	26%	0% (0 of 51)
3	72%	1.3% (1 of 82)
4	97% — one source reproduces this as 94%	43% (13 of 31)
5	100% — all 6 patients died	100% (56 of 56)
6	no patient scored 6; 100% is an extrapolation	not reported

A severity ranking from one hospital in 1998, and what it can bear

The ICH score was built to give intracerebral haemorrhage the kind of standard severity stratification stroke already had, so that case series and trials could be compared. Logistic regression on a retrospective review of every patient presenting with acute intracerebral haemorrhage to one San Francisco hospital in 1997 and 1998 picked out five independent predictors of 30-day mortality — presenting Glasgow Coma Scale, age 80 or over, volume of 30 cm³ or more, intraventricular extension and infratentorial origin — and the points are their rounded regression weights. Five items, a minute at the bedside, still in use nearly three decades later.

The outcomes it reported rise steeply: 0 per cent 30-day mortality at a score of 0, then 13, 26, 72 and 97 per cent, and 100 per cent at 5. Read the denominators before quoting them. The paper’s own anchors are that all 26 patients scoring 0 survived and all six scoring 5 died. Six patients. No patient in the cohort scored 6 at all, so the 100 per cent routinely attached to 6 is extrapolation rather than observation. And when Malinova and colleagues applied the score to 233 patients managed with fibrinolytic therapy two decades later, mortality at scores of 1 to 4 was 0, 0, 1.3 and 43 per cent. The score still separated patients; the absolute numbers had moved almost beyond recognition.

Two of the five items are softer than they look. The GCS item wants the presenting score, which in practice is often recorded after intubation — and an intubated patient has no valid total at all. The volume item wants 30 cm³, measured in the derivation by ABC/2, which overestimates by about 20 per cent on average against planimetry and more in large, irregular or lobar clots. A haematoma near the threshold can score the point or not depending on who measured it, and one point here is a sixth of the range and can separate the 26 per cent stratum from the 97 per cent one. Which leads to what the score cannot be used for. The 2022 AHA/ASA guideline states that baseline severity scales “can be useful to provide an overall measure of hemorrhage severity but should not be used as the sole basis for limiting life-sustaining treatments”, and the reason is circular: the cohorts that produced these figures included patients in whom treatment was withdrawn because the grade was high. A grade is not a diagnosis and a cohort risk is not this patient’s probability: a stratum in which 72 per cent died tells you about that cohort, not which 72 per cent. This page reports what a stratum predicted in a study. It recommends no action. Every threshold here comes from a named cohort, and cohorts differ in case mix, era and treatment; where your unit’s protocol differs, it takes precedence.

Frequently asked questions

What are the five components of the ICH score?

Presenting Glasgow Coma Scale (13 to 15 scores 0, 5 to 12 scores 1, 3 to 4 scores 2), age 80 or over (1), haematoma volume of 30 cm³ or more (1), intraventricular haemorrhage (1) and infratentorial origin (1). The total runs 0 to 6.

What 30-day mortality does each ICH score carry?

In the derivation cohort: 0 per cent at 0, then 13, 26, 72 and 97 per cent at 1 to 4, and 100 per cent at 5. At least one reproduction gives the score-4 figure as 94 rather than 97. No patient in that cohort scored 6, so the 100 per cent quoted for 6 is extrapolated.

Does a high ICH score mean the patient will die?

No. The percentages describe a retrospective single-centre cohort from 1997 and 1998; a later cohort managed with fibrinolytic therapy had mortality of 0, 0, 1.3 and 43 per cent at scores of 1 to 4. The AHA/ASA guideline says baseline severity scales should not be the sole basis for limiting life-sustaining treatment.

How should the haematoma volume be measured for the ICH score?

The derivation used ABC/2, so the 30 cm³ threshold was calibrated against it. ABC/2 overestimates volume by about 20 per cent on average compared with CT planimetry, and more in large, irregular or lobar clots, so a haematoma close to 30 cm³ may or may not score the point.

Is the GCS stratum in the ICH score the same as mild, moderate and severe?

No, and conflating them is a scoring error. The ICH score uses 13 to 15, 5 to 12 and 3 to 4; the conventional bands are 13 to 15, 9 to 12 and 3 to 8. A GCS of 7 is conventionally severe but scores only one ICH point, the same as a GCS of 12.

Related calculators

References

Hemphill JC 3rd, Bonovich DC, Besmertis L, Manley GT, Johnston SC. The ICH score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke. 2001;32(4):891–897.
Malinova V, Iliev B, Mielke D, Rohde V. Intracerebral haemorrhage score allows a reliable prediction of mortality in patients with spontaneous intracerebral haemorrhage managed by fibrinolytic therapy. Cerebrovasc Dis. 2020;48(3-6):165–170.
Greenberg SM, Ziai WC, Cordonnier C, et al. 2022 guideline for the management of patients with spontaneous intracerebral hemorrhage. Stroke. 2022;53(7):e282–e361.
Intracerebral haemorrhage score. Life in the Fast Lane, litfl.com (accessed 7 October 2026).

Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/