In order to predict the normal states of a continuous process with load fluctuations, a nonlinear modeling technique using historical data during normal operations has been proposed. This technique is called DB (database) modeling because historical data during normal operations are accumulated in the database and are used directly for the predictions. The prediction procedure using the DB model is quite simple; it simply searches for similar records to the inputs to the DB model and averages the outputs from the DB model. In advance, it is only necessary to calculate the distances between the records accumulated in the database in order to determine the reference distance used for searching for similar records to the model inputs. The DB model was demonstrated using experiments involving a tank-pipeline system.