Abstract Introduction Anatomical characteristics of patent foramen ovale (PFO) vary a lot, potentially impacting the embolic risk, selection of the device and the final sealing results of the closure procedure. Objective To categorize the different PFO anatomies and evaluate the consequence on the type of device used and the incidence of residual shunt. Methods 624 consecutive patients who underwent PFO percutaneous closure with transoesophageal echocardiography (TEE) guidance were retrospectively reviewed. Four types of PFO were identified once a rigid wire was in place using per-procedural TEE. This new classification was based on the PFO sheathed tunnel’s length and width, presence of concomitant atrial septal aneurysm (ASA) or additional defect (ASD), width of the septum secundum (SS) and aortic enlargement (Figure 1). Type I, or "single PFO" was defined by the absence of ASA or ASD, a tunnel width ≤ 4mm, SS’s width ≤ 10mm and no aortic enlargement (i.e., aortic root diameter <21mm/m², distorted anatomy or scoliosis). Type II or "dominant septal aneurysm" was defined by the presence of an ASA, with a distinction between subtype IIa (long PFO tunnel with length ≥10mm) and subtype IIb (short PFO tunnel with length <10mm). Type III or "modified PFO" was defined by SS’s width > 10 mm, aortic enlargement, or severe scoliosis, and type IV or "multi perforated" was defined by the presence of a concomitant ASD or a multifenestrated septal wall. In case of concomitant criteria, the predominant order was IV, then IIb, then III. For each category, the presence of residual shunt was evaluated after Valsalva manoeuvres with contrast echocardiogram 6 months after the procedure. Results Our population had a median age of 51 [41-59] years and included 43.8% women. The type II PFO was the most frequently encountered, with 42% of the patients (Table 1). A total of 44 (7.1%) patients presented multiple types concomitantly (mostly type II and III). PFO occluders were the most implanted devices overall, particularly with type I and IIb, while ASD occluder were more frequently used with types IV and IIb. Cribriform/Uni™ devices were more frequently used with type IV. Large devices (i.e., left disc’s diameter >25 mm) were predominantly used with type III and IV, but scarcely employed with type I and II. Within type II anatomies, large devices were significantly more frequently used with type IIb. Fluoroscopy duration was significantly higher with type IV. At 6 months ultrasound evaluation, large right-to-left residual shunt was significantly more frequent with type IV anatomy and with patients presenting multiple types concomitantly (18/34 - 40.9%). No significant differences were observed regarding clinical outcomes during follow-up. Conclusion Based on specific anatomical characteristics, four types of PFO may be identified, presenting with increasing procedural complexity, and requesting specific closure devices.