The chloride mass balance (CMB) method for estimating groundwater recharge is economic and effective, provided that the hydrological conditions for its applications are met and the modeling parameters are known. However, modeling parameters such as precipitation and Cl- deposition rates vary temporally, most notably as a result of the climatic changes from late Pleistocene to Holocene. The temporal variability of atmospheric Cl- input and annual precipitation were considered in this study by using a discrete steady state CMB model with different parameters for late Pleistocene and Holocene. Cl- deposition rates, estimated from 36Cl data, were lower in late Pleistocene than Holocene at Yucca Mountain, Nevada, but higher in late Pleistocene than Holocene at Black Mesa, Arizona. Paleoclimate proxies at both Yucca Mountain and Black Mesa point to higher precipitation rates in late Pleistocene than Holocene. The resulting average recharge estimates for Black Mesa are 9 ¿ 5 mm/yr for Holocene and 35 ¿ 22 mm/yr for late Pleistocene. Local recharge rates at Yucca Mountain were estimated from the 36Cl/Cl ratios and Cl- concentrations in perched waters. The estimated recharge for Yucca Mountain is 5 ¿ 1 mm/yr for Holocene and 15 ¿ 5 mm/yr for late Pleistocene. These recharge rates are comparable to results of independent numerical groundwater flow models and watershed-scale infiltration models at Black Mesa and Yucca Mountain, respectively.