In recent clinical gait studies, the use of net energy cost (EC) outcomes has been recommended. This study assessed the comparative reproducibility of gross and net EC of walking in adults and children with locomotion disorders, using a portable system for gas-exchange. Furthermore, methodological considerations were proposed for improving the reproducibility of these outcomes. Assessment of reproducibility was made in adults with polio residuals (n=14) versus healthy adults (n=14), and in children with cerebral palsy (n=13) versus healthy children (n=10). There were four repeated measurement sessions on different days, with inter-session periods of 1 week. All the measurements had been conducted at the outpatient clinic between 2002 and 2004. Based on the Generalizability Theory for data-analysis, it was shown that there was less intra-subject variability in gross EC of walking compared to net EC of walking among adults and children with locomotion disorders, which was apparent from the lower standard error of measurement values. Data-analysis optimisation and study design adjustments were considered and presented in a decision scheme, to demonstrate that the reproducibility of net EC can substantially be increased. The use of gross EC, rather than net EC, in patients with locomotion disorders seems a more reproducible measure of walking efficiency. However, reproducibility of net EC can substantially be improved by careful standardization and using a multiple repetition design, as a result of which this outcome measure becomes suitable for detecting clinically relevant changes at the individual level.