Predicting the effects of land-use changes on water quality is a necessary step in helping policymakers address solutions to nonpoint-source pollution. Watershed water quality models, such as the Soil Water Assessment Tool (SWAT), have been used to model not only these changes, but the hydrology of the system as well. This study compared the ability of SWAT against an analytic element, conjunctive groundwater-surface water model (GFLOW) to model the hydrology in two Iowa watersheds of contrasting hydrology. The goal was to evaluate which approach best simulated the groundwater and surface water hydrology of the watersheds. The South Fork watershed (SFW) and the Walnut Creek (WC)-Squaw Creek (SC) watershed (WCSCW) contain similar geologic materials including till, loess, paleosol, and alluvium units whose hydraulic conductivities (K) range from 10-5 to 10-10 m/s. The SFW is 78,000 ha in area, contains 85 percent row crops and is 80 percent tile drained. The WCSCW is 9,960 ha in area and is characterized by row crop production, prairie restoration, and significantly less tile drainage. Models utilized streamflow data (SWAT, GFLOW) and hydraulic head data from piezometers (GFLOW only) for calibration. In the SFW, a recharge (R) of 4.3 in/yr was able to calibrate the GFLOW model and produced a base flow of 81.2 cfs at USGS gage (05451210), while an R of 0.43 in/yr taken from the calibrated SWAT model was unable to reproduce observed hydraulic heads during the study period and resulted in a base flow of 6.8 cfs at the gage. In contrast, in the WCSCW hydraulic heads could be calibrated based on R values of 3.8 in/yr (WC) and 5.2 in/yr (SC) that are similar in both models. Groundwater discharge from SWAT (6.79 cfs at SQW2) was more similar to base flow from GFLOW (7.54 cfs at SQW2) than in the SFW. Results of this comparison suggest that groundwater recharge values taken from the SWAT-M model are not realistic in tile-drained watersheds such as SFW, but that both models simulate hydrology in the WCSCW. Due to the problem of non-uniqueness in highly-parameterized models such as SWAT, incorporation of tile-drainage into deterministic groundwater models with fewer parameters, such as GFLOW, may ultimately provide a more accurate simulation of the overall hydrology.