Although the third heart sound (S3) is well recognized as an important sign in the evaluation of patients with congestive heart failure, the interobserver variability with its observation needs to be known before general applicability can be determined. Therefore, we determined the agreement among four trained observers on the presence of S3 in 81 hospitalized patients. Agreement between pairs of observers varied between 48 and 73 percent. The kappa statistic, which adjusts for agreement by chance alone, showed that agreement between various observer pairs was moderate (kappa = 0.40-0.50) at best and slight (kappa = 0.10-0.30) at worst. The rate of agreement did not appear to be affected by the time interval between measurements, by the sex of the patient or by a training effect over the time of the study. In conclusion, although S3 may be important as a clinical sign, clinicians cannot agree reliably about whether or not it is present.