The World Health Organization's global programme for the control of acute respiratory infections relies on counting respiratory rate (RR) by observing abdominal and chest movements in order to diagnose pneumonia. However, few studies on the reliability of the observation method have been published. We counted RR simultaneously by observation and auscultation in 100 healthy infants at 1, 2, 4, 6 and 8 weeks of age for 15, 30 and 60 sec, and compared RRs obtained by the two methods. In all the age groups studied, the co-efficients of variation for the RRs recorded by observation or auscultation were similar. The mean RR by observation was higher by 1-3 breaths/min than mean RR by auscultation (p < 0.001). The 95% confidence interval (+/-2 SD) for the difference between RR by the two methods ranged from +5 to -8 breaths/min for RR counted for 1 full minute. Our data support the assumption that observation is as reliable as auscultation for counting RR.