Prediction of groundwater recharge (S) is a very important component of hydrologic modeling. However, such estimates may be expected to be in error. This paper reports a study of the sensitivity of S estimates to errors () in parameters and input data. The investigation was carried out using a soil moisture deficit model (SMDM) by varying the various parameters in the SMDM equation to determine their influence on the predicted recharge amounts. This process was applied to data from two major ecological regions of Nigeria. The results show that the model estimates are highly sensitive to values of precipitation (Pr) and evaptranspiration (Et). Therefore estimation of these parameters is very necessary.