What hemoglobin level is critically low, and when is a transfusion given?

When people search for a critical hemoglobin, they are usually holding one result and asking whether it is an emergency. The honest answer starts by separating three numbers that are often blurred together. One defines how severe an anemia is. One tells a laboratory when it must phone a clinician. One guides a doctor deciding whether to give blood to a patient already in hospital. They come from different bodies and they do not agree, because they are not answering the same question.

Laid side by side, the gaps are the point. A result of 6.5 g/dL is severe anemia by WHO's definition and under the 7 g/dL line at which the AABB recommends considering transfusion in a stable inpatient. Whether the laboratory phones about it depends on the laboratory: by our reading of the survey, it sits above the low limit on at least half of the lists, because the median limit is 6.0, and below it on lists that set the limit as high as 7.0.

The AABB (the Association for the Advancement of Blood and Biotherapies) published its current general guideline in JAMA in November 2023. Its first recommendation reads: for hospitalized adult patients who are hemodynamically stable, the panel recommends a restrictive transfusion strategy considering transfusion when the hemoglobin concentration is less than 7 g/dL. That is a strong recommendation based on moderate certainty evidence. In line with the thresholds used in most trials, it adds that clinicians may choose 7.5 g/dL for patients undergoing cardiac surgery and 8 g/dL for those undergoing orthopedic surgery or those with preexisting cardiovascular disease.

A second recommendation covers hospitalized adults with blood and cancer disorders: again, consider transfusion under 7 g/dL, as a conditional recommendation on low certainty evidence. The guideline also states, as good practice: consider the overall clinical context and alternative therapies to transfusion when deciding about an individual patient.

Notice what the guideline is not. Both adult recommendations are written for hospitalized patients, and the main one for patients who are hemodynamically stable, so it does not describe someone whose blood pressure is falling from active bleeding, and it was not written to tell anyone at home whether to go to hospital.

One group gets a higher line. The MINT trial randomized 3,504 patients with a heart attack and a hemoglobin under 10 g/dL to a restrictive strategy (transfuse at 7 or 8) or a liberal one (transfuse below 10). Death or another heart attack within 30 days occurred in 16.9 percent of the restrictive group and 14.5 percent of the liberal group, a risk ratio of 1.15 with a confidence interval of 0.99 to 1.34. That missed statistical significance, and the investigators wrote that potential harms of the restrictive strategy cannot be excluded.

The AABB's 2025 guideline for acute myocardial infarction, built on a meta-analysis of the randomized trials in heart attack patients, suggests a liberal strategy when hemoglobin is under 10 g/dL, states that a restrictive strategy of 7 to 8 g/dL may result in increased mortality in these patients, and says the decision should rest on the clinical context rather than the hemoglobin alone.

The AABB's 2023 count was 45 adult trials with 20,599 participants comparing a restrictive line, typically 7 to 8 g/dL, with a liberal one of 9 to 10 g/dL. One of them, TRICC, published in 1999, randomized 838 critically ill patients to transfusion below 7 or below 10. Thirty-day mortality was 18.7 percent with the lower threshold and 23.3 percent with the higher one, a difference that was not statistically significant (P = 0.11). The investigators concluded that the restrictive strategy was at least as effective as the liberal one, with the possible exception of patients with acute heart attack and unstable angina.

The 2025 Cochrane review now pools 61 adult trials with 27,639 participants. Restrictive strategies cut the chance of receiving any red cell transfusion by 42 percent and did not change 30-day mortality (risk ratio 1.01, 95 percent confidence interval 0.90 to 1.14, high-certainty evidence). Two clinical groups broke the pattern in opposite directions. In gastrointestinal bleeding, 30-day mortality was lower with the restrictive approach (risk ratio 0.63). In critically ill patients with brain injury, neurological outcomes at 6 to 12 months were better with the liberal approach. Transfusion reactions were uncommon, and more frequent with the liberal strategy.

Federal CLIA rules require US laboratories to list its imminently life-threatening test results, or panic or alert values, in its procedure manual, and to immediately alert the person who ordered the test when a result crosses one. The rules do not say what the values should be. Each lab chooses.

A 2025 survey collected 50 critical value lists from university hospitals, Level 1 trauma centers and other high-performing US hospitals. Ninety-six percent listed a low hemoglobin limit. The median was 6.0 g/dL and the range 5.0 to 7.0, down from a median of 7 g/dL in a 1990 survey, a change the authors found was not statistically significant. The authors flagged the drift: these limits have shifted away from thresholds associated with transfusion need, and shifting them away from severe clinical thresholds may delay detection of life-threatening conditions for some patients.

People search "how long can you live with a hemoglobin of 5". No source we read gives a duration. What exists is a rare and sobering data set: surgical patients who declined transfusion for religious reasons, followed after their operations.

In that study, 300 adults operated on between 1981 and 1994 had a postoperative hemoglobin of 8 g/dL or less. Their mean age was 57.

After adjusting for age, cardiovascular disease and illness severity, the odds of death rose 2.5 times for each 1 g/dL fall below 8. The authors concluded that risk becomes extremely high below 5 to 6 g/dL. These were people recovering from surgery and blood loss decades ago, so the figures describe that setting, not a forecast for anyone else. They are the only outcome data we found for hemoglobin this low without transfusion.

The NHLBI notes that symptoms of anemia may develop quickly or slowly depending on the cause, and that mild anemia may cause no symptoms at all. The 2025 Cochrane authors call for new studies to recognize the clinical context and the limitations of adopting a single hemoglobin threshold. Two people with the same 7.5 can be in very different situations: one with a long, slow iron deficiency who is walking about, and one who is bleeding now.

Make sure the clinician who ordered the test has seen a result under 8 g/dL, because it is severe by definition and the cause has to be found. MedlinePlus advises contacting your provider if you have any symptoms of anemia or unusual bleeding, and if you notice blood or black, tar-like stool. If you think you might be having a heart attack, and chest pain or discomfort is the most common symptom, the NHLBI says to call 9-1-1 even if you are not sure. That applies whatever your hemoglobin is.

Our hemoglobin reference page shows the WHO bands with their source and effective date, and each band has its own page, including severe anemia and moderate anemia. The lab result lookup will place your number in its band. For a result that is low but not in this range, our guides to normal hemoglobin levels and the ferritin range cover why WHO's cutoff flags some healthy people and why iron stores are the next number to ask for.

A result under 8 g/dL is severe anemia by WHO's definition, so make sure the clinician who ordered it has seen it. MedlinePlus advises contacting your provider for any symptoms of anemia, such as shortness of breath, dizziness or fainting, or for unusual bleeding. Useful questions: how fast did this fall, what is the cause, does my heart history change the threshold, and what is the plan if it falls further.