The ground‐motion variability sigma is a fundamental component in probabilistic seismic‐hazard assessment because it controls the hazard level at very low probabilities of exceedance. So far, most of the analyses based on empirical ground‐motion prediction equations do not consider any distance dependency of sigma. This study aims to analyze the potential distance dependency of ground‐motion variability, especially in the near‐field region, where the variability is poorly constrained due to the lack of available records. We, therefore, investigate the distance dependency of sigma by performing numerical simulations of ground motion for some strike‐slip events. Synthetic velocity seismograms (up to 3 Hz) have been generated from a suite of finite‐source rupture models of past earthquakes. Green’s functions were calculated for a 1D velocity structure using a discrete wavenumber technique (Bouchon, 1981). The within‐event component of the ground‐motion variability was then evaluated from the synthetic data as a function of distance. The simulations reveal that the within‐event component of the ground motion shows a distance dependency, subject to the rupture type. For bilateral ruptures, the variability tends to increase with distance. On the contrary, in case of unilateral events, the variability decreases with distance.