Population studies of fish depend upon correct age estimates. However ageing from otoliths is often a subjective activity, based on experience. To produce more objective and reproducible results, image analysis systems for semi-automatic otolith reading are being developed. We propose determinig the daily growth and annual structures in each otolith by means of Fourier analysis of the luminous signal, after filtering the high frequencies by means of a moving threshold. A new approach using spatio-frequential wavelet analysis is discussed for otolith age determination. The different readings of each otolith are standardised and weighted to obtain a combined age determination and vector of increment widths.