AB: The accurate prediction of near-source strong-ground motions in future earthquakes is a major challenge for seismologists and earthquake engineers. Such a prediction is however difficult, partly because ground motions are highly variable while the reasons for such a variability are basically ignored. In particular, though source effects are likely an important factor, the major properties of the earthquake source that control the ground motions are not clearly defined. Manighetti et al. (2007) have suggested that some of the large earthquake source properties are intimately controlled by intrinsic properties of the long-term faults broken by the earthquakes, such as the fault structural maturity. Following that suggestion, we examine the dependency of the strong ground motion variability on the long-term fault maturity. We analyze the near-field ground motions recorded at rock sites for 30 large (M 5.6-7.8) crustal earthquakes of various mechanisms. On the other hand, the structural maturity of the broken faults is determined and defined into three classes (mature, intermediate, immature) based on the combined knowledge of the age, slip rate, cumulative slip and length of the faults. The ground motions are determined in the period range 0-2 seconds and compared to the empirical attenuation equation of Boore et al. (1997). Residuals between observed and predicted ground motions are calculated following the methods of Kagawa et al. (2004; residuals not weighted by the number of records per earthquake) and of Spudich et al. (1999; residuals weighted by the number of records per earthquakes). Whatever the method, the strong motions produced by earthquakes on immature and mature faults are found to be larger and lower, respectively, than those predicted from the empirical law. Strong motions on immature faults result to be systematically larger by a factor of about 1.15 than those on mature faults. When earthquakes are distinguished from their focal mechanism, a same order difference is observed, with ruptures on reverse faults producing larger strong motions than earthquakes on strike-slip faults. Yet, when only strike-slip earthquakes are considered, a large difference in strong motion amplitude (averaging a factor 1.1) is still observed between immature and mature faults. By contrast, whether they produce surface breaks or not, earthquakes on faults of similar structural maturity produce ground motions of similar amplitude. We conclude that the strong ground motions generated in the near-field by earthquakes occurring on immature faults are higher than the strong motions generated by earthquakes on mature faults. The effect of fault structural maturity is of the same magnitude than that classically attributed to focal mechanism. The later effect may actually be only apparent, simply resulting from the maturity control. The structural maturity of long-term faults is thus an important parameter that should be considered in seismic hazard assessment.