The authors present a new model for the ageing of a lead-acid battery which is based on the initial model of Shepherd. The proposed model allows to predict temporal variations of the Shepherd coefficients and to control the deterioration of the battery parameters and performances. The model validation has been realised by the recursive least square (RLS) algorithm by using long-term measurements under several solicitations. This study will improve the storage section of stand-alone photovoltaic systems and reduce overloads and deep discharges.