AbstractThe basic computational procedure of a newly developed seismic performance test méthod is presented and examined in this paper. This method uses a direct step‐by‐step integration technique to compute the displacement response of a test specimen subjected to a numerically specified seismic excitation record, utilizing the non‐linear restoring forces actually developed by the specimen during the experiment. Due to the limitations of the experimental procedure and conditions, special analytical assumptions and numerical methods must be adopted. The adequacy of such analytical techniques is evaluated and possible computational errors are identified. An improved numerical integration algorithm is also proposed for this specific application. It has been found that the method can be a reliable and powerful experimental tool provided appropriate analytical and numerical criteria, as discussed in this paper, are observed.