Arterial blood gas analysis is used to measure the pH and the partial pressures of oxygen and carbon dioxide in arterial blood. The investigation is relatively easy to perform and yields information that can guide the management of acute and chronic illnesses.This information indicates a patient's acid-base balance, the effectiveness of their gas exchange and the state of their ventilatory control. Interpretation of an arterial blood gas result should not be done without considering the clinical findings.The results change as the body compensates for the underlying problem. Factors relating to sampling technique, specimen processing and environment may also influence the results.


Arterial blood gas analysis is a common investigation in emergency departments and intensive care units for monitoring patients with acute respiratory failure. It also has some application in general practice, such as assessing the need for domiciliary oxygen therapy in patients with chronic obstructive pulmonary disease. An arterial blood gas result can help in the assessment of a patient's gas exchange, ventilatory control and acid–base balance. However, the investigation does not give a diagnosis and should not be used as a screening test. It is imperative that the results are considered in the context of the patient's symptoms.

While non-invasive monitoring of pulmonary function, such as pulse oximetry, is simple, effective and increasingly widely used, pulse oximetry is no substitute for arterial blood gas analysis. Pulse oximetry is solely a measure of oxygen saturation and gives no indication about blood pH, carbon dioxide or bicarbonate concentrations.

Arterial puncture

Blood is usually withdrawn from the radial artery as it is easy to palpate and has a good collateral supply. The patient's arm is placed palm-up on a flat surface, with the wrist dorsiflexed at 45°. A towel may be placed under the wrist for support. The puncture site should be cleaned with alcohol or iodine, and a local anaesthetic (such as 2% lignocaine) should be infiltrated. Local anaesthetic makes arterial puncture less painful for the patient and does not increase the difficulty of the procedure.1 The radial artery should be palpated for a pulse, and a pre-heparinised syringe with a 23 or 25 gauge needle should be inserted at an angle just distal to the palpated pulse (Fig. 1). A small quantity of blood is sufficient. After the puncture, sterile gauze should be placed firmly over the site and direct pressure applied for several minutes to obtain haemostasis. If repeated arterial blood gas analysis is required, it is advisable to use a different site (such as the other radial artery) or insert an arterial line.

To ensure accuracy, it is important to deliver the sample for analysis promptly. If there is any delay in processing the sample, the blood can be stored on ice for approximately 30 minutes with little effect on the accuracy of the results.

Complications of arterial puncture are infrequent. They include prolonged bleeding, infection, thrombosis or arteriospasm.

Fig. 1

Performing an arterial puncture
arterial puncture

Interpreting a blood gas result

The automated analysers measure the pH and the partial pressures of oxygen (PaO2) and carbon dioxide (PaCO2) in arterial blood. Bicarbonate (HCO3ˉ) is also calculated (Box 1). These measurements should be considered with the patient's clinical features (Table 1).

Box 1

Reference ranges for arterial blood gases





Base excess

7.35 – 7.45

80 – 100* mmHg

35 – 45 mmHg

22 – 26 mmol/L

–2 – +2 mmol/L

10.6 – 13.3 kPa

4.7 – 6.0 kPa

Reference ranges for venous blood gases





7.32 – 7.43

25 – 40 mmHg

41 – 50 mmHg

23 – 27 mmol/L

* age and altitude dependent (see text)
Kilopascals: to convert pressures to kPa, divide mmHg by 7.5

Table 1
Correlating arterial blood gas results with clinical features

Metabolic imbalances Respiratory imbalances

Metabolic acidosis Metabolic alkalosis Respiratory acidosis Respiratory alkalosis
PaCO2 N (uncompensated)
↓ (compensated)
N (uncompensated)
↑ (compensated)
HCO3ˉ N (uncompensated)
↑ (compensated)
N (uncompensated)
↓ (compensated)
Base excess N/↑ N/↓
Clinical features Kussmaul-type breathing (deeper, faster respiration), shock, coma Paraesthesia, tetany, weakness Acute: air hunger, disorientation
Chronic: hypoventilation, hypoxia, cyanosis
Acute: hyperventilation, paraesthesia, light-headedness
Chronic: hyperventilation, latent tetany
Common causes

With raised anion gap: diabetic ketoacidosis, lactic acidosis, poisons (e.g. ethylene glycol), drug overdoses (paracetamol, aspirin, isoniazid, alcohol)

With normal anion gap: diarrhoea, secretory adenomas, ammonium chloride poisoning, interstitial nephritis, renal tubular acidosis, acetazolamide administration

Vomiting, prolonged therapy with potassium-wasting diuretics or steroids, Cushing's disease, ingestion/overdose of sodium bicarbonate (e.g. antacids) Hypoventilation
chronic lung disease with CO2retention, e.g. chronic obstructive pulmonary disease, respiratory depression from drugs (e.g. opioids, sedatives), severe asthma, pulmonary oedema
Hyperventilation anxiety, pain, febrile illness, hypoxia, pulmonary embolism, pregnancy, sepsis

N = within normal range ↑ = increased ↓ = decreased


The pH determines the presence of acidaemia or alkalaemia. If the body has compensated for the disorder, the pH may be in the normal range.


The PaCO2reflects the state of alveolar ventilation. An elevated PaCO2 reflects alveolar hypoventilation, whereas a decreased

PaCO2 reflects alveolar hyperventilation. Acute changes in PaCO2 will alter the pH. As a general rule, a low pH with a high PaCO2 suggests a respiratory acidosis, while a low pH with a low PaCO2 suggests a metabolic acidosis.

There is a delayed response of PaCO2 to an acute change. Increases in PaCO2 occur relatively slowly, as the body's overall CO2 stores are very large (approximately 20 L) and the volume of CO2 generated by metabolism (200 mL/min) makes little overall difference. For instance, during a breath-hold, the PaCO2 rises at a rate of only 2–3 mmHg per minute, hence patients with a very high PaCO2 usually have a long-standing disorder. Accordingly, even when treated the PaCO2 may take a long time to return to normal.

The state of arterial blood oxygenation is determined by the PaO2. This reflects gas exchange in the lungs and normally the PaO2 decreases with age. This is due to decreased elastic recoil in the lungs in the elderly, thereby yielding a greater ventilation-perfusion mismatch. The expected PaO2 when breathing air at sea level can be calculated with the equation PaO2 = 100 – (age x 0.25). Consequently, a PaO2 of 75 mmHg, which may be of concern in a young person, is usually unremarkable in an 85-year-old.


A PaO2 that is less than expected indicates hypoxaemia. This can result from hypoventilation or a mismatch of ventilation and perfusion. If alveolar ventilation is adequate (that is, PaCO2 is normal), then the hypoxaemia is almost certainly caused by a ventilation-perfusion disturbance. The nature of the hypoxaemia can be further assessed by the difference between the alveolar and arterial oxygen tensions.

The alveolar–arterial oxygen tension difference

If an arterial blood gas result shows hypoxaemia (low PaO2) and inadequate alveolar ventilation (high PaCO2), it must be determined whether the hypoxaemia is related to hypoventilation, or is secondary to a disturbance in ventilation-perfusion, or both. This is assessed by calculating the difference between the alveolar (PAO2) and arterial (PaO2) oxygen tensions (see Box 2).

The alveolar–arterial difference, or gradient, can be estimated only if the oxygen fraction of inspired air (FiO2, usually 0.21 on room air), barometric pressure and water vapour pressure are known. A normal reference range is 5–15 mmHg. The difference, expressed as P(A–a)O2, increases with age, cigarette smoking and increasing FiO2. An expected P(A–a)O2 can be calculated using the formula P(A–a)O2 = 3 + (0.21 x patient's age).

All causes of hypoxaemia, apart from hypoventilation, increase the alveolar-arterial difference. In a patient breathing room air, a P(A–a)O2 greater than 15 mmHg suggests a ventilation-perfusion mismatch related to disease of the airways, lung parenchyma or pulmonary vasculature. However, the result is non-specific in defining the actual pathology and again the patient's clinical features are essential for diagnosis.

Box 2

The alveolar–arterial oxygen gradient

P(A-a)O2 = PAO2 – PaO2

PaO2 = arterial oxygen tension

PAO2 = alveolar oxygen tension

PAO2 = FiO2(PB - PH2O) - 1.2(PaCO2)

FiO2 = oxygen fraction in inspired air

PB = barometric pressure (760 mmHg at sea level)

PH2O = water vapour tension (47 mmHg at 37° C)

Normal value


Bicarbonate is a weak base that is regulated by the kidneys as part of acid–base homeostasis. The HCO3ˉ measured in arterial blood reflects the metabolic component of arterial blood. Together, CO2and HCO3ˉ act as metabolic and respiratory buffers respectively. They are related via the equation:

H2O + CO2 H2CO3 HCO3ˉ + H+

Compensatory changes

For any disturbance of gas tensions in arterial blood, a compensatory system exists to maintain homeostasis. In a metabolic disorder, where HCO3ˉ may be retained or excreted by the kidneys, respiratory compensation can occur almost immediately to alter the rate and depth of ventilation to retain or remove CO2. This occurs due to the exquisite sensitivity of chemoreceptors in the medulla to carbonic acid (H2CO3) or H+. Renal compensation in response to a respiratory disorder takes much longer, sometimes between three and five days, to retain or remove HCO3ˉ as required.

As a general rule, when compensation is present the arterial blood gas result shows two imbalances – derangement of both HCO3ˉ and PaCO2. A clue to which imbalance is the primary disturbance is obtained from the pH. If pH is leaning toward acidosis or alkalosis, then the parameter that matches the pH trend (that is, is increased or decreased corresponding to pH) is the primary problem and the other is due to compensation.

The base excess

The metabolic component of the acid–base balance is reflected in the base excess. This is a calculated value derived from blood pH and PaCO2. It is defined as the amount of acid required to restore a litre of blood to its normal pH at a PaCO2 of 40 mmHg. The base excess increases in metabolic alkalosis and decreases (or becomes more negative) in metabolic acidosis, but its utility in interpreting blood gas results is controversial.

While the base excess may give some idea of the metabolic nature of a disorder, it may also confuse the interpretation. The alkalaemia or acidaemia may be primary or secondary to respiratory acidosis or alkalosis. The base excess does not take into account the appropriateness of the metabolic response for any given disorder, thus limiting its utility when interpreting results.

Anion gap

The anion gap assists with the diagnosis of metabolic acidosis (Box 3). This difference between the concentrations of measured anions and cations increases with dehydration and decreases with hypoalbuminaemia. The gap also widens if there is an increase in the concentration of unmeasured anions such as ketones and lactate.

Box 3

The anion gap concept

  • the anion gap is an artificial concept that may indicate the cause of a metabolic acidosis
  • it represents the disparity between the major measured plasma cations (sodium and potassium) and the anions (chloride and bicarbonate)
  • when calculating the anion gap, potassium is usually omitted from the calculation thus: Gap = Na+ + (Cl- + HCO3-)
  • the anion gap is normally between 8 and 16 mmol/L
  • a raised anion gap indicates an increased concentration of lactate, ketones or renal acids and is seen in starvation and uraemia
  • a raised anion gap is seen in overdoses of paracetamol, salicylates, methanol or ethylene glycol
  • a normal anion gap is seen if a metabolic acidosis is due to diarrhoea or urinary loss of bicarbonate

Factors influencing blood gas results

A number of sampling and environmental factors may affect the result of the analysis. Delayed processing of the sample may yield a falsely low PaO2, as the delay allows leucocytes to consume oxygen. This can be avoided by prompt transport of the sample on ice.

Air bubbles introduced when performing the arterial puncture can also cause a falsely high PaO2 and a falsely low PaCO2.2 This can be avoided by gently removing air bubbles within the specimen immediately after collection without agitating the sample.

Body temperature can also affect arterial blood gas tensions. This is relevant in febrile or hypothermic patients, so body temperature should be recorded at the time of collection.3

Mixed acid–base disorders

It is possible to have a mixed respiratory and metabolic disorder that makes interpretation of an arterial blood gas result difficult. As a general rule, when a normal pH is accompanied by an abnormal PaCO2or HCO3ˉ then a mixed metabolic-respiratory disorder exists.Table 2 provides some common clinical examples of mixed respiratory and metabolic disturbances, and Fig. 2 and Fig. 3 are algorithms for the consideration of primary and mixed acid–base disorders.4

Table 2
Examples of mixed acid–base disorders

Mixed metabolic/respiratory disturbance Example

Respiratory acidosis and metabolic acidosis

A patient with acute pulmonary oedema after an acute myocardial infarct
Mechanism: poor cardiac circulation (causing a lactic acidosis – metabolic acidosis) with concurrent poor alveolar ventilation (due to pulmonary oedema) – causing CO2retention and a concomitant respiratory acidosis
Respiratory alkalosis and metabolic alkalosis A patient with hepatic cirrhosis who is given diuretics
Mechanism: patients with hepatic cirrhosis can experience the phenomenon of the hepatopulmonary syndrome where the major symptom is dyspnoea (causing a respiratory alkalosis), while diuretics can cause a decrease in blood volume, which stimulates the renin-angiotensin-aldosterone system, increasing the exchange between Na+ and K+ or H+ at the distal tubule, resulting in an increase in bicarbonate concentration and a metabolic alkalosis
Respiratory acidosis and metabolic alkalosis A patient with long-standing chronic obstructive pulmonary disease who is given diuretics for concomitant heart failure
Mechanism: long-standing air flow limitation may cause chronic hypercapnia and respiratory acidosis via impaired CO2excretion, while diuretics can cause a decrease in blood volume, which stimulates the renin-angiotensin-aldosterone system, increasing the exchange between Na+ and K+ or H+ at the distal tubule, resulting in an increase in bicarbonate concentration and a metabolic alkalosis
Respiratory alkalosis and metabolic acidosis A patient with chronic renal failure who begins to hyperventilate secondary to anxiety
Mechanism: chronic renal failure causes a metabolic acidosis by uraemia and failure to excrete acids while the respiratory alkalosis results from blowing off excess CO2due to alveolar hyperventilation
Metabolic acidosis and metabolic alkalosis A patient with chronic renal failure who suffers from severe intractable vomiting
Mechanism: chronic renal failure causes a metabolic acidosis by uraemia and failure to excrete acids while a concurrent metabolic alkalosis results from the depletion in the body stores of H+ and Cl2 through vomiting
Fig. 2
Interpreting acidaemia on an arterial blood gas result

Image adapted from reference 4 with permission
Fig. 3
Interpreting alkalaemia on an arterial blood gas result
Image adapted from reference 4 with permission

Limitations of blood gas analysis

The blood gas analysis cannot yield a specific diagnosis. A patient with asthma may have similar values to another patient with pneumonia. Alternatively, a patient with chronic obstructive pulmonary disease and respiratory failure may have similar results to a patient with pulmonary oedema.

The analysis does not reflect the degree to which an abnormality actually affects a patient. A low PaO2 does not necessarily indicate tissue hypoxia, nor does a normal PaO2 indicate adequate tissue oxygenation. Oxygen utilisation is influenced by other factors such as regional blood flow, haemoglobin affinity for oxygen and cardiac output.

Blood gas analysis cannot be used as a screening test for early pulmonary disease. Severe disease may be present before significant changes are seen in blood gases.

Venous blood gases

It is easier to obtain a venous sample than an arterial sample. In some situations analysis of venous blood can provide enough information to assist in clinical decisions. In general, the pH, CO2and HCO3ˉ values are similar in venous and arterial blood (Box 1). The main difference is the partial pressure of oxygen in venous blood is less than half that of arterial blood. Venous blood should not therefore be used to assess oxygenation.


Measuring arterial blood gases can be a useful adjunct to the assessment of patients with either acute or chronic diseases. The results show if the patient is acidaemic or alkalaemic and whether the cause is likely to have a respiratory or metabolic component. The PaCO2 reflects alveolar ventilation and the PaO2 reflects the oxygenation of arterial blood. When combined with a patient's clinical features, blood gas analysis can facilitate diagnosis and management.

Conflict of interest: none declared

If you prescribe subscribe

Further Reading

Martin L. All you really need to know to interpret arterial blood gases. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 1999.


  1. Lightowler JV, Elliot MW. Local anaesthetic infiltration prior to arterial puncture for blood gas analysis: a survey of current practice and a randomised double blind placebo controlled trial. J R Coll Physicians Lond 1997;31:645-6. [R] .
  2. Harsten A, Berg B, Inerot S, Muth L. Importance of correct handling of samples for the results of blood gas analysis. Acta Anaesthesiol Scand 1988;32:365-8. .
  3. Williams AJ. ABC of oxygen: assessing and interpreting arterial blood gases and acid-base balance. BMJ 1998;317:1213-6. .
  4. Drage S. Acid base balance. Update 13. 2001. World Federation of Societies of Anaesthesiologists. .