Although the existence of a time delay worsens the diagnosis accuracy, most research efforts on fault diagnosis have not considered the time delay. The hybrid fault diagnosis method that is based on the signed digraph (SDG) and the partial least-squares (PLS) has the advantage of improving diagnosis resolution and accuracy, compared to those of previous qualitative methods, and is capable of enhancing the ability to diagnose multiple-fault [Lee et al., Ind. Eng. Chem. Res. 2003, 42, 6145-6154]. The current research modified this method to include the information of the time delay between process variables, and the time delay can be obtained through the open-loop responses of set-point changes. The modified method is applied for the fault diagnosis of the pulp mill process, which produces pulp from wood chips. It is one of the biggest processes in the fault diagnosis area, and it will be a new benchmark process to evaluate the fault diagnosis method. Also, its units, such as the storage tank and the bleaching tower, have a very large time delay. Dynamic principal component analysis (DPCA)-based fault diagnosis was used to get a reference of the diagnosis performance comparison. Although fault detection of the proposed method is frequently slower than DPCA, it gives more accurate results. Also, it demonstrates much-better diagnosis capabilities, compared to the original hybrid method.