In spring 1999, repeated larval surveys were undertaken by IFREMER from March to July on a major spawning ground of anchovy off Gironde estuary. In September, a juvenile survey was performed in collaboration with AZTI. The paper describes the methodology for reading the larval and juvenile otoliths as well as for reconstructing the correspondence in space and time between juveniles and larvae using an 3D hydrodynamic model. The thin otoliths (sagittae) of the larvae were read directly using a light microscope. The thicker otoliths (sagittae) of the juveniles had to be polished before reading. To read the narrow increments near the nucleus (first weeks after mouth opening), it was necessary to use a greater magnification (x1250) than for the overall reading (x312.5). Because the otolith increases along preferential axes after larval stage, measurements were always performed in a similar manner relative to these axes. Virtual buoys were released in the hydrodynamic model on a regular grid at dates compatible with the estimated juvenile birthdates. The trajectories that arrived near juvenile sampling locations at dates compatible with sampled juvenile ages were retained and their origin was thus evidenced. Some of the juveniles sampled in September could have originated from the French spawning ground off the Gironde estuary in June and July. The comparison of the survived juvenile growth rates during their larval period with the otolith growth rate of the June and July larvae supported the idea of growth-selective survival : the survived September juveniles showed faster growth than the pool of June larvae they were estimated to originate from. Variations in otolith growth pattern also suggest a higher juvenile growth over the shelf break than in oceanic waters.