A simple but accurate prediction method for the performance of a commercial spiral wound forward osmosis (FO) module was developed. The method includes a fundamental model based on the internal and external concentration polarization (ICP and ECP) in the FO process, the module experiments with sodium chloride, and a fitting process to increase the accuracy of the model. Comparison between the ICP/ECP modeling results and the experimental data reveals that trans-membrane pressure is a key parameter to explain the difference between the modeling and experimental results. The draw solution (DS) channel heights in the module could be shrunk by trans-membrane pressure, and this hypothesis is strongly supported by a statistical analysis for the differential pressure and the accuracy of a modified model accounting for the effect of trans-membrane pressure. The modified model with trans-membrane pressure and other hydrodynamic factors as fitting variables predicts the performance of the spiral wound FO module accurately. The number of experimental data for the fitting process (i.e., trained data) to obtain a reliable prediction model was found to be around 30, and the modified model turns out to work reasonably well to predict not only the trained data but also the untrained data.