Effective arterial elastance (E(a)) was originally defined as the end-systolic pressure (ESP)/stroke volume (SV) ratio of the left ventricle (LV). E(a) combined with LV contractility (E(max)), E(a)/E(max), proved to be powerful in analyzing the ventriculo-arterial coupling of normal and failing hearts in regular beats. However, E(a) sensitively changes with LV E(max), preload, and afterload widely changing among irregular beats. This has discouraged the use of E(a) during arrhythmia. However, we hypothesized that E(a) could serve as the effective afterload (not always arterial) elastance against ventricular ejection under arrhythmia. We tested this hypothesis by analyzing beat-to-beat changes in E(a) of irregular beats during electrically induced atrial fibrillation (AF) in normal canine in situ hearts. We newly found that during AF in each heart: 1) E(a) changed widely among irregular beats and became markedly high in weak beats with small SVs; 2) E(a) and E(a)/E(max) distributed non-normally with large skewness but 1/E(a) distributed more normally; 3) 1/E(a) correlated closely with end-diastolic volume, E(max) and preceding beat intervals; and 4) the reciprocal of mean 1/E(a) closely correlated with mean ESP/mean SV. These results support our hypothesis that E(a) can serve as the effective afterload elastance against ventricular ejection on a per-beat basis during AF. E(a)/E(max) can also quantify the ventriculo-afterload (not arterial) coupling on a per-beat basis. This study, however, warns that mean E(a) and mean E(a)/E(max) of irregular beats cannot necessarily represent their averages during AF.