In this paper, the condition of a full-scale concrete bridge subjected to strong earthquakes is evaluated using vibration-based analyses. A new computational toolkit is developed in MATLAB environment for damage identification and long-term monitoring. Two types of parametric and non-parametric analysis methods are carried out on monitoring data. Moreover, a dynamic performance index is proposed based on an AutoRegressive Moving Average with eXogenous excitation (ARMAX) model. This index utilises the response predicted from an ARMAX model to evaluate bridge behaviour during strong earthquakes. Based on the results, a minor but permanent drop of 0.05 Hz in natural frequency of the first transverse and vertical modes is observed after the first strong earthquake. Also, a significant drop in frequency of the first transverse mode is observed during the two strong earthquakes. The results of the index show that the bridge did not follow linear behaviour during the two strong earthquakes as expected from a linear system. A close to flag-shaped force-displacement relationship is also observed during the first strong earthquake that can be an indication of nonlinearity in bridge behaviour. The analysis results illustrate the efficiency of the new monitoring platform for long-term monitoring and management of large datasets.