Reliable measurements of resting metabolic rate (RMR) in humans are crucial to detect meaningful changes in its status. Past research suggests that low intra-measurement variation of respiratory gases (steady state) is necessary to achieve valid measurements on critically ill patients (1). However, it is unclear how stability of continuously assessed gases may affect RMR reliability in healthy individuals (2,3). Therefore, this study aimed at defining the effect of data analysis methods determining lowest coefficient of variation within indirect calorimetry measurements on the reliability of between and within-day RMR. To assess this, healthy participants (n=26) completed three measurements of RMR (two on the same day, and two on consecutive days) using a ventilated hood linked to the Vyntus CPX system. Each measurement consisted of 30 min of individuals resting in supine position in standardised conditions, with the first 10 min used for adaptation, providing the subsequent 20 min of net collected data per measurement. Within this 20 min period we determined RMRs for the intervals of lowest coefficient of variation (CV) for VO2, VCO2, and respiratory exchange ratio of 5 and 10 min duration, using data sampling rate (smoothing) of 10, 20 and 30 sec as well the whole 20 min of data collected and the first and second 10 min intervals, providing a total of 21 methods of interval determination for RMR. Between-day RMRs showed no systematic bias, but intervals ≥ 10 min showed better typical error of measurement (TEM; 3.7-4.6%) and intraclass correlation (ICC; 0.92-0.95) than intervals of 5 min (TEM, 4.8-6.1%; ICC, 0.86-0.92). There was no relationship between average within-measurement CV and between day CV (Pearson R2 ≤ 0.009, p ≥ 0.2). Whole period and the first 10 min interval showed no systematic bias, a nearly perfect correlation of 0.92 (0.84-0.97; 95%CL) and 0.93 (0.85-0.97) respectively and a small TEM of 4.5 (3.5-6.2) and 4.3% (3.3-5.9%) respectively. There was a nearly perfect correlation between the 1st 10 min interval and the whole 20 min period (R2 = 0.991, p < 0.0001). Within-day RMR showed a lower reliability, evidenced by lower correlations between measurements (ICC, 0.85-0.89) and higher typical error (TEM, 6.8-8%). Systematic bias seemed negligible, but in some cases the second measurement was lower than the first one (effect sizes ~0.20; 95%CL ~0.45-0.05). In conclusion, criteria for determining lowest variation (CV) during measurements seemed to have no meaningful effects on the reliability of measurements, but longer periods provided more reliable between day RMR data. Additionally, there was no relationship between within-measurement CV and RMR reliability. Ten and 20′ of data collected under standardised conditions provide reliable RMR measurements without need of considering the CV of data analysed.