Aims Functional mitral regurgitation (FMR) is a consequence of mitral annular enlargement, leaflet tethering and reduced co-aptation. The importance of the left atrium (LA) as a cause of mitral regurgitation (MR) is less clear. We applied a co-aptation index using three-dimensional (3D) transoesophageal echocardiography to FMR and MR secondary to LA dilatation (atrial mitral regurgitation, AMR). Methods and results Seventy-two patients underwent comprehensive 3D echo studies: FMR (n = 19); AMR (n = 33); and 20 controls. We recorded: LV size and function; LA dimensions; mitral annular area (MVA); and leaflet area in early and late systole - MVA fractional change was defined: (MVA late systole - MVA early systole)/MVA late systole x 100; the co-aptation index was defined: (leaflet area early systole - leaflet area late systole)/leaflet area early systole x 100. Despite normal LV size and function in AMR, MVA was increased similarly to FMR (AMR 12.86 cm(2) vs. FMR 12.33 cm(2), P = ns; both P < 0.01 vs. controls 8.83 cm(2)), and MVA fractional change similarly reduced (AMR 5.1 vs. FMR 6.3; P = ns; both P < 0.001 vs. controls 14.6). The co-aptation index was reduced in both MR groups (FMR 6.6 vs. AMR 7.0, P = ns; both P < 0.001 vs. controls 19.6). After multivariate analysis, the co-aptation index (chi(2) = 41.2) and MVA fractional change (chi(2) = 22.1) remained the strongest predictors of MR (both P 0.001 for the model). A co-aptation index of <= 13 was 96 sensitive and 90 specific for the presence of MR. Conclusion LA dilatation leads to MVA enlargement, reduced leaflet co-aptation and MR even without LV dilatation. A co-aptation index describes this in vivo. This work provides insights into the mechanism of AMR.