The in vitro cavitation thresholds and locations were studied on ten different heart valve designs. The valves were mounted in a mock circulation flow loop which simulated a cardiovascular system. All the tests were run at 70 beats per minute with a cardiac output varying between 2 l/min and 6 l/min in increments of 1 l/min. In vitro cavitation phenomena generated at the closing instant of mechanical heart valves were captured using a video photographic technique. Cavitation locations and intensity on different valve designs were analyzed from the cavitation images recorded on a video tape. When cavitation occurs on a bileaflet valve, it can occur in the same localized area of the leaflet from cycle to cycle thus producing a cumulative effect. In a single disc valve, the free rotation of the valve disc during operation provides a means of distributing a localized cavitation activity over an ever changing disc surface. Thus any cavitation-induced damage on the disc surface can be reduced or eliminated even though a single disc valve may have a lower cavitation threshold. Cavitation locations and thresholds are primarily a function of valve design. Smaller size valves have higher cavitation thresholds than larger ones. The cavitation thresholds of all the valves tested were above the physiological left ventricular maximum dp/dt at rest. If in vivo cavitation occurs under some extreme conditions, this study suggests possible locations on mechanical heart valves which could be examined for traces of cavitation activity.