The inland water level plays an essential role in water balance management. This paper presents methodology to retrieve water level from radar altimeter data including Cryosat-2 and Jason-2 over lakes in Western Africa. The idea is to combine both Level-1 and Level-2 data to retrieve water level more precisely and automatically over small lakes. From Level-1 data, the waveform is analyzed for data filtering and the possibility of water level retrieval. Then an iteration method is proposed to retrieve the height from Level-2 altimeter data with a strict water mask. The preliminary results of water level time series, derived from Level-2 data, in Lake Volta and Lake Kainji are presented in this paper, and the results are compared with the data from DAHITI (Database for Hydrological Time Series of Inland Waters). The standard deviation is controlled below 0.3 meters.