This study investigated anatomic and pathophysiologic variables that may determine which patent foramen ovale (PFO) are associated with cerebrovascular accidents (CVAs). Anatomic features of a PFO have been identified as risk factors that predispose certain people to cryptogenic strokes (strokes of unknown cause). However, potential pathophysiologic variables that can determine the pressure gradient between left and right atria, which could influence the right-to-left shunt through a PFO, have not been examined. A retrospective study included 78 consecutive patients in whom PFOs were detected during routine transesophageal echocardiography examination. Group I included 36 patients with CVAs of unknown cause (cryptogenic stroke). Group II included 42 patients without CVAs whose PFOs were incidental findings. Anatomic features measured included separation and overlap between septum primum and septum secundum, interatrial septal motion, and the relative size of the right-to-left shunt with peripheral saline solution contrast injections. Pathophysiologic variables considered were those that could cause elevated left atrial pressure, thereby minimizing the right-to-left shunt.Patients with a clinical neurologic event (group I) had a larger right-to-left shunt volume of contrast bubbles than did patients with asymptomatic PFO (group II; P =.004). The size of the overlap between septum primum and septum secundum was less in patients from group I as compared with patients from group II (7.5 +/- 3.4 mm versus 9.9 +/- 6.0 mm; P =.026). However, other anatomic features of PFO that are determinants of the "opening" were not significantly different between the 2 groups. No statistically significant difference was found in the distance of the separation between septum primum and septum secundum (2.5 +/- 2.0 mm versus 1.9 +/- 1.6 mm; P = not significant). The prevalence of interatrial septal aneurysm was also similar between the 2 groups (28% versus 21%; P = not significant). However, the prevalence of variables that could potentially raise left atrial pressure was greater in patients without CVA as compared with those with a CVA (48% versus 14%; P =.01). In our study, anatomic features alone did not determine interatrial shunt size and pathophysiologic variables that could raise left atrial pressures did differentiate between patients with a PFO with a CVA/transient ischemic attack and those without it. Thus, both anatomic and pathophysiologic mechanisms should be considered in determination of the potential clinical significance of a PFO.