Measurement of respiratory rate (RR) is essential in the evaluation of respiratory disorders. However, the variability in RR measurement in adults has never been adequately assessed. Respiratory rate was measured twice in 245 patients; the two measurements were performed by the same observer in 137 patients, by different observers in 58 patients and simultaneously by different observers in 50 patients. The mean (SD) difference between the first and second measurements was 0.03 (3); 95% limits of agreement-4.86-4.94 breaths min(-1), -5.7-5.7 breaths min(-1), and -4.2 to 4.4 breaths min(-1) for the same observer, different observer and simultaneous observer groups, respectively. The difference in RR measurements did not vary with RR. In conclusions on average, there is very good agreement between observers in RR measurement. Inter-observer variability may account for a difference of up to 6 breaths min(-1). This is relevant when applying clinical prediction rules based on threshold RR values.