Combined use of the tritium/helium 3 (3H/3He) dating technique and particle-tracking analysis can improve flow-model calibration. As shown at two sites in the Great Miami buried-valley aquifer in southwestern Ohio, the combined use of 3H/3He age dating and particle tracking led to a lower mean absolute error between measured heads and simulated heads than in the original calibrated models and/or between simulated travel times and 3H/3He ages. Apparent groundwater ages were obtained for water samples collected from 44 wells at two locations where previously constructed finite difference models of groundwater flow were available (Mound Plant and Wright-Patterson Air Force Base (WPAFB)). The two-layer Mound Plant model covers 11 km2 within the buried-valley aquifer. The WPAFB model has three layers and covers 262 km2 within the buried-valley aquifer and adjacent bedrock uplands. Sampled wells were chosen along flow paths determined from potentiometric maps or particle-tracking analyses. Water samples were collected at various depths within the aquifer. In the Mound Plant area, samples used for comparison of 3H/3He ages with simulated travel times were from wells completed in the uppermost model layer. Simulated travel times agreed well with 3H/3He ages. The mean absolute error (MAE) was 3.5 years. Agreement in ages at WPAFB decreased with increasing depth in the system. The MAEs were 1.63, 17.2, and 255 years for model layers 1, 2, and 3, respectively. Discrepancies between the simulated travel times and 3H/3He ages were assumed to be due to improper conceptualization or incorrect parameterization of the flow models. Selected conceptual and parameter modifications to the models resulted in improved agreement between 3H/3He ages and simulated travel times and between measured and simulated heads and flows. ¿ 1998 American Geophysical Union