Aortic valve stenosis (AS) in patients >75 years of age is a challenge for diagnosis and management of every day clinical routine. Therefore, this clinical follow-up study aims to investigate predictors of death in patients with advanced stages of AS. In a single-center study, all patients (n = 157) with primary conservatively treated severe AS (mean age 78 ± 6 years) were included. All patients had initially refused aortic valve replacement (AVR). During a median follow-up of 2.6 years (quartiles 1.7, 3.8), 62 patients with severe AS switched to AVR and 95 remained conservatively treated (no AVR). Routine clinical data were assessed together with conventional echocardiography including the measurement of longitudinal wall function and deformation (mitral ring displacement and longitudinal strain and strain rate imaging). The end points were all-cause and cardiac death. During follow-up, cardiac death occurred in 49% in no-AVR group. In a Cox regression analysis, New York Heart Association functional class, valvuloarterial impedance, stroke volume, longitudinal strain and strain rate, and mitral annular displacement identified an increased risk of all-cause death (hazard ratio [HR] for mitral annular displacement 15.9, 95% confidence interval [CI] 6.24 to 40.86, p <0.001, positive predictive value 91%). In contrast, ejection fraction or EuroSCORE was not predictive (ejection fraction: HR 1.3, 95% CI 0.82 to 2.33, p = 0.25; EuroSCORE: HR 1.1, 95% CI 0.64 to 2.02, p = 0.64). Furthermore, in multivariate regression analysis, only longitudinal mitral annular displacement and longitudinal strain rate was a significant predictor of all-cause and cardiac death risk. These data show that prognosis in elderly patients with AS is determined by symptoms, hemodynamics, and particularly by cardiac long-axis function. Thus, for risk stratification, a comprehensive assessment of cardiac function including the measurement of longitudinal mitral annular displacement should be considered.