Quantifying the role of atrial ectopic foci in the initiation and the maintenance of atrial fibrillation (AF) is a relevant clinical problem. This study aims at assessing the relation between the activity level of an ectopic focus and the complexity of the resulting AF dynamics. Eight different episodes of focal AF were simulated in a 3D model of human atria based on Courtemanche atrial kinetics. Focal sources were introduced as a group of cells firing at a fixed rate. To create various AF dynamics, the episodes differed by the location of their focal source. Clinically relevant sites were selected in the left and right atrium. Their activity level was quantified by the number of propagating focal fronts originating from the source within a 10-s time window. The dynamic complexity was evaluated through the time course of the atrial equivalent dipole orientation. The 3rd eigenvalue (smallest) of the covariance matrix of the normalized dipole vector was used to characterize the spread of spatial distribution of the atrial dipole. Different degrees of dynamic complexity were observed in the 8 episodes, from broad wavelets/stable macroreentries to multiple wavelets/spirals. The number of focal fronts within 10 s ranged from 21 to 43 (30.9±7.2). The 3rd eigenvalue varied between 0.13 and 0.23 (0.19±0.03). The maximum of the dipole’s spatial distribution consistently pointed toward the location of the source when the dynamic complexity was low. A negative correlation (-0.93) was found between the 3rd eigenvalue and the number of focal fronts. An outlier was observed, which was characterized by the lowest dynamic complexity (3rd eigenvalue of 0.13) in spite of a non-extreme number of focal fronts (28) caused by a stable macroreentry masking the focal source. Analyzing the spatial distribution of the atrial dipole, possibly estimated from non-invasive data, could provide information about the activity level of atrial ectopic foci and the complexity of AF dynamics.