Background: We sought to evaluate the effect of left ventricular (LV) systolic function and age on determining the degree of correlation between plasma NT-proBNP levels and LV filling pressures (LVFP) and their optimal values for predicting elevated LVFP. Methods: Total 359 consecutive patients referred to coronary angiography were studied. Patients with valvular heart disease, creatinine level > 1.5 mg/dL and atrial fibrillation were excluded. LV pre-A pressures (LVPAP) were measured by fluid-filled catheters and compared with intraventricular plasma NT-proBNP levels, simultaneously. Echo-Doppler derived LV ejection fraction (EF), left atrial volume index (LAVI) and LV mass index (LVMI) were measured. Results: All the patients were divided into four groups as I (n=190, EF���50% and age<65), II (n=100, EF���50% and age���65), III (n=41, EF<50% and age <65) and IV (n=28, EF<50% and age���65). In each groups, log-transformed NT-proBNP levels correlated significantly with LVPAP when adjusted with LAVI, LVEF, age and LVMI. However correlations were better in LV dysfunctional group (r=0.605 in group III, p<0.01 and r=0.583 in group IV, p<0.01) than in others (r=0.466 for group I, p<0.05 and r=0.534 in group II, p<0.05). The optimal cutoff NT-proBNP level to predict LVPAP>15 mmHg varies with aging and LV systolic function (group III=238.6 pg/ml as sensitivity 77.2% and specificity 77.8%, group IV= 511.8 pg/ml as 88.9% and 82.4% , p<0.01) but its diagnostic accuracy was lower in patients with preserved LV systolic function (group I= 86.6 pg/ml, as sensitivity 70.0% and specificity 63.3%, p <0.05, group II=181.9 pg/ml as 76.9% and 65.9%, p <0.01). Conclusion: Plasma NT-proBNP level can be a predictor for elevated LVFP, but its cutoff value and degree of correlations vary according to LV systolic function and age.