This study is intended to investigate the feasibility of bridge health monitoring using a linear system parameter of a time series model identified from traffic-induced vibration data of bridges through a moving vehicle experiment on scaled model bridges. In order to detect possible abnormality of bridges, this study adopted a parameter from autoregressive (AR) coefficients. Consideration goes into diagnosis of the bridge from pattern change of identified system parameters due to damage. Mahalanobis-Taguchi system (MTS) is applied to make a decision on bridge health condition. Observations demonstrate feasibility of structural diagnosis of bridges from the identified system parameter of the vehicle-bridge interactive system.