This study aimed to (1) estimate the reliability of 36 centre of pressure (COP) summary measures in healthy participants and (2) identify the main sources of variability in order to estimate the most appropriate measurement strategies to improve reliability. Twelve healthy males performed, on two separate days, eight 1-min trials of quiet standing on a force platform in two conditions [eyes open (EO) and eyes closed (EC)]. The generalizability theory was used as a framework to estimate the magnitude of the different variance components (Subject, Trial, Day and all interactions) and the reliability of the measures corresponding to various simulations of measurement strategies. Reliability of the COP summary measures was poor to moderate. Intra-class correlation coefficients were generally higher with EO (mean: 0.46, range: 0.03-0.76) than with EC (mean: 0.41, range 0.02-0.72) across all summary measures. The majority of the variance was attributed to Subject (2%-76%), Subject x Day (0%-24%) and Subject x Day x Trial (16%-79%) variance components depending on the summary measure and condition. The reliability could be improved more efficiently by averaging measurements between-days than by increasing the number of trials during 1 day. For the majority of the summary measures, acceptable reliability can be achieved when at least seven or more trials are averaged during the same testing day.