Flowmeters are necessary devices for the operation of chemical plants, but foreign matter accumulation in a pipe or wear of a part such as the edge of an orifice hole may give rise to measurement errors. A faulty flowmeter may cause a loss of operational stability, so it is important to detect faults early and repair or replace the meter. In order to diagnose faults of a sensor such as a flowmeter, a method using deviations from linear relationship established between indicated values has been proposed. However, from the viewpoint of automization of fault diagnosis for practical use, several points remain to be examined concerning a method of fault detection and diagnosis that is robust to measurement noise. Therefore, this study proposes a method for realizing automatic fault diagnosis by examining these points. The effectiveness of the proposed method was demonstrated by automatic fault diagnosis of a flowmeter by running a dynamic simulation of a vinyl acetate monomer process.