Background: Noninvasive diagnosis of exercise-induced elevation of left ventricular filling pressure is difficult and remains unsatisfactory. The aim of this study was to assess the accuracy of the ratio of early diastolic transmitral (E) to mitral annular (e) velocity and to determine new parameters or parameter combinations with the ability to predict exercise-induced left atrial pressure (LAP) elevation. Methods and results: Eighty patients with paroxysmal atrial fibrillation (AF) referred for catheter AF ablation underwent simultaneous exercise echocardiography and direct invasive LAP measurements, as well as a resting and postexercise biomarker analysis. Exercise E/e >=8.85 predicted exercise LAP >=20 mm Hg with 61.5% sensitivity and 88.9% specificity (area under the curve [AUC], 0.76). Of all of the individual parameters tested, the best prediction was achieved with exercise E/s (s=peak systolic mitral annular velocity) >=8.75 (sensitivity, 88.5%; specificity, 64.8%; positive predictive value, 54.8%; negative predictive value, 92.1%; AUC, 0.84). However, the combination of exercise E/A (A = late diastolic transmitral flow velocity) >=1.22 + exercise E/e >=8.85 + exercise s<=11.05 cm/s provided the most precise prediction of exercise LAP elevation (sensitivity, 84.6%; specificity, 79.6%; positive predictive value, 66.7%; negative predictive value, 91.5%; AUC, 0.90). Conclusions: Exercise E/e, when used as a sole parameter, was not sufficiently reliable to predict exercise-induced elevation of LAP. The application of a multivariate-adjusted combination of parameters appeared to be the preferable approach for the noninvasive prediction of exercise LAP elevation.