The North-West area of the Northern Kanto groundwater basin is characterized by its complicated multi-aquifer structure and thus it is difficult to apply inverse analysis to estimate all parameters. Therefore, the aquifer parameter identification has been carried out by the combination of geostatistic kriging, trial-and-error identification method and inverse analysis. First, the initial aquifers' heads and prior transmissivities were estimated by geostatistic kriging. Second, the leakage factors of the confining layers were calibrated by a trial-and-error method which has been automatically carried out. Third, the storage coefficient was identified by trial-and-error method. Finally, the prescribed boundary flux intensities and transmissivity of the aquifers were estimated by an inverse analysis by means of Fletcher-Reeves conjugate gradient method. This sequential procedure seemed to be suitable for this practical case. Since the trial-and-error method and geostatistic kriging are widely well known, they will not be described in detail in this paper, but only briefly. This paper stresses on the use of combination of different parameter identification methods and mainly focuses on the utilization of the Fletcher-Reeves gradient method to solve a groundwater flow inverse analysis problem.