The structure and connectivity of European seabass (Dicentrarchus labrax) populations remain poorly known and ecological evidence is missing to support the current delineation between the northern (southern North Sea, English Channel and Celtic Sea) and southern French stocks (Bay of Biscay). Adult spawning site fidelity and natal homing were analysed by coupling Data Storage Tag (DST) information and otolith microchemistry of recaptured fish to investigate, within the study area, the population structure and connectivity in European seabass. Trajectory reconstructions inferred from DST data were used to assign a spawning area (English Channel or Bay of Biscay) to each spawning winter record. In addition, otolith composition (Mg, P, Mn, Zn, Sr, Ba and δ18O) was measured in both larvae and adults otolith increments corresponding to a winter spawning event. We built a training dataset using coupled spawning area assignments and otolith elemental signatures (Mg, P, Mn, Zn, Sr and Ba) for winters with DST data. The training dataset was used to calibrate a Random Forest model and assign spawning areas based on otolith winter signatures outside the DST recording period. Results revealed that 64% of the seabass expressed spawning site fidelity. We also found a geographical gradient of site fidelity, with the highest proportions of spawning site fidelity found in seabass tagged at the northern and southern limits of the studied area. Significant ontogenetic effects were observed for trace elements and δ18O with ratios significantly lower in the larval stage than in the adult stage. These biases and the variability across cohorts prevented us to use the assignment model fitted on adults to study natal homing. At the larval stage, the analysis of spatio-temporal effects on otolith trace elements did not reveal any significant difference between spawning areas. However, the patterns of difference were similar for larval and adult Zn, Sr and Ba between the two spawning areas, suggesting a homing behaviour.