The repeatability of energy-expenditure measurements were studied in five children and four adults without disabilities using the Cosmed K4 (high technology). The ability to detect change in measurements was compared between this instrument and the Physiological Cost Index (PCI; low technology). The results of repeatability (95% range) for oxygen cost were 13.1% in children and 13% in adults. In contrast, the SD of PCI was 6 to 72% of the mean in adults and wider in children (91%; 95% range). The validity of PCI as an outcome measure was questioned. In addition, 177 children with motor disability were prospectively studied using the Cosmed K4. Previous experience with the Cosmed K2 (intermediate technology) helped to develop a practical and repeatable protocol for testing children with disability using the Cosmed K4. The protocol commenced with 5 minutes of rest to achieve baseline values of heart rate and oxygen consumption, followed by 10 minutes of continuous walking at a self-selected speed on a 10-metre level oval walking track. The test concluded with 5 minutes of rest to monitor the return to baseline values. Ninety-one percent of the children with disability quickly reached a steady-state of oxygen consumption and carbon-dioxide production. The carbon-dioxide sensor in the Cosmed K4 has enabled a new group of severely involved children with cerebral palsy (9%) to be defined. These children have been termed 'physiologically marginal ambulators'.