Nine different mechanical mitral heart valves were chosen in order to study cavitation dynamics in detail in an in vitro flow system simulating a single event of mitral valve closure. The transvalvular pressure (ventricular minus atrial pressure) rise rate averaged during the valve closing period was used as an index of the loading rate. A series of photographs in the vicinity of the inflow surface of the valve were attempted during the bubble appearance period using a stroboscopic visualization technique. The in vitro study revealed three sources of cavitation initiation in the design of the mechanical heart valves tested: stop (seat stop or seating lip), inflow strut, and clearance (gap formed between the occluder and the housing or between the two occluders in the closed position). Among these, the occluder stop design was the most critical to cavitation since all valves having the stop at the edge of the major orifice area showed a higher intensity of cavitation and threshold loading rates below the estimated normal physiological value. The analysis of bubble locations and dynamics led us to propose that the fluid squeezing effect between the occluder and the stop in the housing and the streamline contraction effect along the clearance are factors responsible for cavitation incipience.