Groundwater recharge and water balance estimation in the Akaki catchment of the Upper Awash Basin, central Ethiopia, is challenging due to hydrological variability and modeling uncertainties. This study quantified groundwater recharge and other water balance components in the Akaki catchment and its sub-catchments—Legedadi and Gefersa—using the WetSpass-M model. Input data included rainfall, temperature, wind speed, and two types of potential evapotranspiration (PET): FAO-Penman-Monteith and GLEAM, used interchangeably. Spatial layers such as land use/land cover (LULC), soil type, slope, and groundwater depth were derived from Landsat 8, ISDASoil, ALOS DEM, and well observations. Streamflow data were used for model calibration, validation, and performance evaluation. Groundwater recharge in Akaki was estimated at 248 mm using FAO-Penman-Monteith PET, and slightly increased to 256 mm with GLEAM. In Legedadi, recharge rose from 244 mm to 262 mm, while in Gefersa it declined from 323 mm to 307 mm. Akaki's water balance components under FAO-Penman-Monteith were 43 % actual evapotranspiration (AET), 36 % runoff, and 21 % recharge. GLEAM PET reduced AET and increased recharge in Akaki and Legedadi, but increased AET and reduced recharge in Gefersa. These adjustments lowered the water balance error in Akaki (−0.72) and Legedadi (−1.62), but slightly increased it in Gefersa (−1.72). Model performance was strong in Akaki (NSE 0.94, R² 0.95, KGE 0.95), moderate in Legedadi (NSE 0.84, KGE 0.70), and good in Gefersa (NSE 0.93, R² 0.90, KGE 0.86). However, dry-season discharge and baseflow mismatches indicate the need for improved PET estimation, higher-resolution data, and refined calibration.