Predicting pesticide fate with a reasonable degree of precision using mathematical models requires a good choice of parameter values. When experimentally derived values are not readily available, or need to be measured individually for each compound through systematic laboratory experiments, the process not only becomes too time-consuming, but also may not yield reliable parameters due to the uncertainties encountered with laboratory measurements. The inverse modelling technique has therefore become an important tool in recent times, allowing calibration of models against experimental data and thus alleviating the lack of exactness, reproducibility, and objectivity often associated with laboratory-derived data and trial–error simulation. In this study, we used the inverse modelling package PEST interfaced with the GLEAMS, HYDRUS-1D, and LEACHM models to derive field-based mobility and degradation parameters for a selected number of pesticides, along with a bromide tracer, applied to 2 contrasting field sites in the south island of New Zealand. Given the broad range in soil properties at both sites and the climatic conditions (one drier than the other) used in testing, the models performed well. Based on the performance of the models, they can be ranked in the order LEACHM > HYDRUS-1D > GLEAMS. Bromacil appeared to be the most mobile among the compounds at both sites as well as having a greater persistency, followed by hexazinone and terbuthylazine, based on their optimised K_oc values at the sites. For bromacil, median optimised K_oc values were 22 and 35 mL/g at Nelson and Southland sites, respectively (compared with a best available value of 14 mL/g). However, T_1/2 values for bromacil were much lower than the available best literature value of 207 days. The median K_oc values for terbuthylazine were 114 and 87 mL/g at Nelson and Southland sites, respectively. These values were much lower than the best available literature value of 220 mL/g, and fall below the literature range (162–278 mL/g for terbuthylazine).