A predictive model is presented for estimating the peak inelastic oscillator displacements (S-d,S-ie) from peak ground velocity (PGV). The proposed model accounts for the variation of S-d,S-ie for bilinear hysteretic behavior under constant ductility (mu) and normalized lateral strength ratio (R) associated with postyield stiffness ratios of alpha = 0 and 5%. The regression coefficients are based on a ground-motion database that contains dense-to-stiff soil site recordings at distances of up to 30 km from the causative fault. The moment magnitude (M) range of the database is 5.2 <= M <= 7.6 and the ground motions do not exhibit pulse-dominant signals. Confined to the limitations imposed by the ground-motion database, the model can estimate S-d,S-ie by employing the PGV predictions obtained from the attenuation relationships (ground-motion prediction equations). In this way, the influence of important seismological parameters can be incorporated to the variation of S-d,S-ie in a fairly rationale manner. This feature of the predictive model advocates its implementation in the probabilistic seismic hazard analysis that employs scalar ground-motion intensity indices. Various case studies are presented to show the consistent estimations of S-d,S-ie by the proposed model. The error propagation in the S-d,S-ie estimations is also discussed when the proposed model is associated with attenuation relationships. Copyright (C) 2008 John Wiley & Sons, Ltd.