We are interested in the stellar content in the central region of active galaxies. In this aim, we perform stellar population synthesis of the integrated starlight in the Optical and NIR using a mathematical method (Pelat, 1997, MNRAS 284, 365) dedicated to the determination of the best solution (among the many local solutions) in the stellar population synthesis problem which is a complex inverse problem. A new development of the code allows the determination of the error zone around the solution, expected from the observational uncertainties. In the next sections, we present this very powerful method and results obtained from a sample of AGN.