In this study, a batch-type, nanoscale, zero-valent iron (nZVI) process was used to treat trichloroethylene (TCE) wastewater. Variations of oxidation-reduction potential (ORP) and pH in the reactor were monitored online for use in developing the control model of the nZVI process. After the addition of nZVI to the reactor, the pH value increased rapidly, from 5.0-6.0 to around 8.5-9.5, whereas the ORP decreased dramatically, from around 300 mV to -700 to -800 mV. The use of an nZVI dose of 1.5 g/L to treat an influent with TCE concentration of 50 mg/L resulted in a TCE removal efficiency of 94%. The kinetics of the degradation of TCE by nZVI followed the pseudo-first-order model with the rate constant (k) ranging from 0.02 to 0.13. Two models, i.e., a multiple regression model and an artificial neural network (ANN) model, were used to develop the control model to predict the TCE removal efficiencies using four operational parameters, i.e., initial TCE concentration, nZVI dose, ORP, and pH. Both the regression model and the ANN model performed precise prediction results for the TCE removal efficiencies, with correlation coefficients (R2) of about 0.87 and 0.98, respectively; thus, the combined use of the regression model and the ANN model, along with monitoring ORP and pH, has great potential for controlling the nZVI process for TCE removal. Finally, another 16 runs of experiments were conducted to evaluate the control models, and the ANN model provided very precise prediction results for TCE removal, with an R2 value of 0.98.