Transcatheter aortic valve implantation (TAVI) is an alternative to surgical aortic valve replacement in very high-risk patients with severe aortic stenosis. The present multicenter, retrospective study investigates the accuracy and calibration properties of the EuroSCORE and the age, serum creatinine, and ejection fraction (ACEF) score in selecting a population of patients to be referred to TAVI. The study includes 1053 surgical and 211 transcatheter procedures. The operative mortality rate within the surgical group was 2%. The EuroSCORE and the ACEF score had similar levels of accuracy; the ACEF score was well calibrated and the EuroSCORE overestimated the mortality risk. The observed mortality rate within the transcatheter group was 10.4%. Cut-off values corresponding to a mortality rate of 10% were 26 for the logistic EuroSCORE and 2.5 for the ACEF score: both the EuroSCORE and the ACEF score may be used to refer patients to TAVI. However, they do not consider a number of 'extreme' risk conditions that may justify a transcatheter procedure even in absence of an overall elevated risk score. These risk conditions should be included in a specific risk model for referring patients for TAVI.