In spite of the current interest in and clinical application of intravascular ultrasound imaging, there is very little published information on the limitations of this modality. In vitro studies on nine phantom vessels (diameter 4.76 to 12.7 mm) and 11 human arteries (diameter 3.7 to 6.2 mm) were performed to assess the potential sources of error of diagnostic intravascular ultrasound imaging. The effects of (1) blood flow velocity, (2) temperature, (3) eccentric, noncentral catheter placement, (4) alteration of the angle of incidence by 30 degrees, and (5) the effect of imaging in different mediums--saline solution, blood, and electrode gel--were studied. Variations in blood flow velocity (from 10 to 300 ml/min) and temperature (from 22 degrees C to 37 degrees C) resulted in a < 2% change in the lumen area measured by intravascular ultrasound imaging catheters. Eccentric catheter location had little effect on phantom or human arterial lumen shape or area when imaging was performed with optimized catheters. However, with used catheters circular lumens appeared elliptical with an eccentric index for phantoms from 0.88 to 1.15, (P < 0.05), and for human arteries from 0.88 to 1.11 (P < 0.05). The area ranged from 89% to 112% (P < 0.05) in phantoms and from 90% to 110% in human arteries compared with the lumen areas measured with a central catheter position (control). A 30-degree alteration in the angle of incidence resulted in 16.3% +/- 5.5% increase in lumen area for phantoms and 14.2% +/- 8.6% for human arteries in vitro. Ultrasonic-measured wall thickness of human vessels correlated closely with the actual measured thickness (r = 0.93) when a central catheter position was used. The wall thickness measured during adjacent (< 0.2 mm) and far-wall positioning (1.9 mm) of the catheter correlated closely (r = 0.96), but the far wall thickness with a 30-degree angle of incidence resulted in a 10.6% increase from control. Studies in saline solution resulted in significantly different measures of lumen area compared with imaging in blood. Compared with images recorded in blood, images in saline solution were 7.6% to 8.2% larger and 3.9% to 7.2% smaller in gel.(ABSTRACT TRUNCATED AT 400 WORDS)