This study examines the reproducibility and variability of pulsed wave Doppler versus continuous wave Doppler ultrasound indexes of left ventricular filling in cardiac allograft recipients and in normal subjects. The following indexes were studied: isovolumic relaxation time, pressure half-time, peak early mitral flow velocity, and peak mitral flow velocity after atrial systole. Intraobserver and interobserver variability were assessed by regression analysis. Individual components of variance (subject, reader, beat, day, and tracing) were estimated in a subset of five patients and five normal subjects, and estimated total variance defined for each group. Temporal (day-to-day) variability for 95% confidence was estimated for these patients and for normal subjects. Temporal variability in the group from which the subsets were drawn was measured from absolute and percent change in values on two occasions. Estimated and observed 95% confidence limits were compared. Intersubject variability was the largest component of variance in both transplant recipients and in normal subjects. For all indexes in transplant recipients (in the absence of rejection) and normal subjects, observed absolute mean differences (+/- 2 standard deviations) between values from recordings taken on two different days were larger than the 95% confidence limits estimated from the components of variance analysis. The observed 95% limits for transplant recipients versus normal subjects were as follows: isovolumic relaxation time, 20 msec versus 6 msec; pressure half-time, 16 msec versus 9 msec; peak early mitral flow velocity, 32 cm per second versus 17 cm per second; and peak mitral flow velocity after atrial systole, 28 cm per second versus 10 cm per second.(ABSTRACT TRUNCATED AT 250 WORDS)