Actual evapotranspiration/evaporation is essential in proper determination of potential recharge of semi-arid regions. Eight potential evapotranspiration methods with dual crop coefficient methodology were used to estimate actual evaporation (E-a) values from bare soil. Method comparison with measured E-a values using lysimeter data for the hydrological years 2011/12 and 2012/13 identified better performance for the FAO-56-Penman-Monteith (NRMSE < 9%), Hargreaves-Samani (NRMSE < 13%), Priestley-Taylor (NRMSE < 15%) and Makkink (NRMSE < 14%) methods. Data of 2011/12, best E-a estimates and a water balance approach were used to calibrate six potential recharge models with linear, quadratic, exponential, cubic and power functions. Calibrated models were validated in simultaneous estimation of potential recharge and soil moisture using independent data for 2012/13. While models with linear/quadratic functions produced the weakest potential recharge estimates (Q > 10%), the remaining models were acceptable. Soil moisture was acceptably simulated utilizing six models and four best evaporation methods (NRMSE < 20%). However, combination of the FAO-56-Penman-Monteith model with an exponential function produced the best E-a values, soil moisture and potential recharge.