Calibration of flow models with field data is complicated by the nonlinear dependency of the unsaturated flow parameters on the water content. Combining predictions using independent models, called multimodel prediction, is state-of-the-art technique. The objective of this study was to compare different methods of multimodel simulation of the field soil water regime using 19 pedotransfer functions (PTFs) with HYDRUS-1D. Different methods of combining the simulation results from the 19 individual models were tested: using only the best model, using equal weights, using regressing measured values to the results of the individual models, using singular-value decomposition (SVD) in the regression, using Bayesian model averaging, and using weights derived from Akaike criteria. Experimental data at five depths along a 6-m transect in a loamy soil were used to calibrate the water flow model. The SVD multimodel was the best method, with an accuracy of about 0.01 m3 m–3 at the 35-cm depth and about 0.005 m3 m–3 at greater depths for 30 d of monitoring and 13 mo of testing. This indicates that multimodeling in combination with monitoring of the soil water regime can be a viable approach to simulating water flow in the vadose zone.