This study examined the test-retest reliability of oxygen cost (ml·kg^-1·min^-1) during level walking in individuals with acquired brain injury (ABI). Ten individuals with ABI (5 men, 5 women) (Traumatic brain injury, 1, central pontine myelinolysis, 1, stroke 8) and 21 healthy controls (11 men, 10 women). Measurements of gross and net (walking minus resting) oxygen consumption (ml·kg^-1·min^-1), and oxygen cost (ml·kg^-1·min^-1) during level walking at self-selected speeds. Measurements were taken on two occasions within one week. Oxygen cost was significantly lower (p < 0.05) in individuals with ABI on the second test versus the first test. Percentage variability in oxygen cost from test to re-test ranged from 14.7 to 17.3% in the control group and from 17.4 to 20.8% in the brain injury group. Clinical populations may demonstrate a significant decrease in oxygen cost between testing occasions. Individuals require at least one period of familiarisation if oxygen cost is used as an outcome measure during level walking in clinical groups. The amount of familiarisation has yet to be investigated in individuals with ABI.