To assess the accuracy and reproducibly of cardiovascular magnetic resonance (CMR) in the measurement of the aortic annulus and in process of valve sizing as compared to intra-operative sizing, cardiovascular computed tomography (CCT) and transesophageal echocardiography (TEE). Retrospective study on 42 patients who underwent aortic valve replacement from September 2010 to September 2015, with available records of pre surgery annulus assessment by CMR, CCT and TEE and of peri-operative assessment. In CCT and CMR, the annular plane was considered a virtual ring formed by the lowest hinge points of the valvular attachments to the aorta. In TEE the annulus was measured at the base of leaflet insertion in the mid-esophageal long-axis view using the X-plane technique. Two double-blinded operators performed the assessments for each imaging technique. Intra-operative evaluation was performed using Hegar dilators. Continuous variables were studied with within-subject ANOVA, Bland–Altman (BA) plots, Wilcoxon’s and Friedman’s tests; trends were explored with scatter plots. Categorical variables were studied with Fisher’s exact test. The intra- and inter-operator reliability was satisfying. There were no significant differences between the annulus dimensions measured by CMR and either one of the three references. Valve sizing for CoreValve by CMR had the same good agreement with CCT and TEE, with a 78 % match rate; for SAPIEN XT the agreement was slightly better (82 %) for CCT than for TEE (66 %). MR performs well when compared to the surgical reference of intra-operative sizing and stands up to the level of the most used imaging references (CCT and TEE).