Transcatheter aortic valve implantation (TAVI) is associated with a higher risk of neurological events for both the transfemoral and transapical approach than surgical valve replacement. Cerebral magnetic resonance imaging has revealed more new, albeit clinically silent lesions from procedural embolization, yet the main source and predominant procedural step of emboli remain unclear. Eighty-three patients underwent transfemoral (Medtronic CoreValve MCV(TF), n=32; Edwards Sapien ES(TF), n=26) and transapical (ES(TA): n=25) TAVI. Serial transcranial Doppler examinations before, during, and 3 months after TAVI were used to identify high-intensity transient signals (HITS) as a surrogate for microembolization. Procedural HITS were detected in all patients, predominantly during manipulation of the calcified aortic valve while stent valves were being positioned and implanted. The balloon-expandable ES prosthesis caused significantly more HITS (mean 95% CI) during positioning (ES(TF), 259.9 184.8-334.9; ES(TA), 206.1162.5-249.7; MCV(TF), 78.5 25.3-131.6; P<0.001) and the self-expandable MCV prosthesis during implantation (MCV(TF), 397.1 302.1-492.2; ES(TF), 88.2 70.2-106.3; ES(TA), 110.7 82.0-139.3; P<0.001). Overall, there were no significant differences between transfemoral and transapical TAVI or between the MCV and ES prostheses. No HITS were detected at baseline or 3-month follow-up. There was 1 major procedural stroke that resulted in death and 1 minor procedural stroke with full recovery at 3-month follow-up in the MCV group. Procedural HITS were detected by transcranial Doppler in all patients. Although no difference was observed between the transfemoral and the transapical approach with the balloon-expandable ES stent valve, transfemoral TAVI with the self-expandable MCV prosthesis resulted in the greatest number of HITS, predominantly during implantation.