Although the S3 gallop sound has long been used clinically as an indicator of left ventricular systolic dysfunction, the mechanism responsible for its production remains controversial. The same sound is often found in young healthy individuals, and whether a similar mechanism is responsible is also unknown. The relationship of the S3 gallop sound to the dynamics of left ventricular filling was compared in 18 healthy young triathletes and 15 older subjects with cardiac disease. Twenty healthy normal subjects without an S3 were included as controls. Phonocardiographic, two-dimensional echocardiographic, and Doppler echocardiographic analysis of left ventricular inflow were evaluated. The S3 in both groups always occurred close to peak early filling velocity (E), during early flow deceleration. Mean E deceleration rate was higher in the subjects with S3 (726 +/- 153 cm/sec2 in the triathletes and 819 +/- 274 cm/sec2 in those with cardiac disease) than in control subjects (563 +/- 131 cm/sec2, p less than 0.001 in both cases). Ten triathletes underwent examination both before and immediately after 30 degrees head-up tilt. E deceleration rate dropped significantly with head-up tilt (720 +/- 137 vs 590 +/- 174 cm/sec2, p less than 0.01), while concurrently the S3 disappeared or was diminished in amplitude. Similar changes were seen in subjects with cardiac disease. We conclude that both the "pathologic" and "physiologic" S3 are related to abnormally rapid deceleration of early diastolic left ventricular inflow. Although the presence of the S3 is not dependent on the state of left ventricular systolic function, diastolic filling is characterized by a predominance of early inflow with a rapid flow deceleration rate.