This study simulates synthetic ground motion time histories tailored to specified earthquake scenarios in Italy, leveraging the ITA18 strong motion database. The proposed approach is based on a modulated and filtered white-noise process, which captures the evolving intensity, predominant frequency, and bandwidth of ground motion time series. Model parameters are obtained by fitting the stochastic model to a subset of accelerograms from the ITA18 dataset. Using these fitted parameters as outputs, vectorized ground motion prediction equations (GMPEs) are developed based on earthquake scenario input variables, such as earthquake magnitude, faulting mechanism, source-to-site distance, and site shear-wave velocity. The GMPEs capture the statistics of the model parameters, including their means, variability, and correlations. For any given earthquake scenario, ensembles of synthetic acceleration, velocity, and displacement time series are generated using the predicted model parameter sets. The resulting motions statistically reproduce the main features of recorded ground motions, such as Arias intensity, peak ground velocity, duration, as well as elastic and inelastic response spectra. This simulation framework supports performance-based earthquake engineering in Italy by enabling site- and scenario-specific ground motion generation, particularly in regions with limited observational data.