The input for the seismic risk analysis can be expressed with a description of "groundshaking scenarios", or with probabilistic maps of perhaps relevant parameters. The probabilistic approach, unavoidably based upon rough assumptions and models (e.g. recurrence and attenuation laws), can be misleading, as it cannot take into account, with satisfactory accuracy, some of the most important aspects like rupture process, directivity and site effects. This is evidenced by the comparison of recent recordings with the values predicted by the probabilistic methods. We prefer a scenario-based, deterministic approach in view of the limited seismological data, of the local irregularity of the occurrence of strong earthquakes, and of the multiscale seismicity model, that is capable to reconcile two apparently conflicting ideas: the Characteristic Earthquake concept and the Self Organized Criticality paradigm. Where the numerical modeling is successfully compared with records, the synthetic seismograms permit the microzoning, based upon a set of possible scenario earthquakes. Where no recordings are available the synthetic signals can be used to estimate the ground motion without having to wait for a strong earthquake to occur (pre-disaster microzonation). In both cases the use of modeling is necessary since the so-called local site effects can be strongly dependent upon the properties of the seismic source and can be properly defined only by means of envelopes. The joint use of reliable synthetic signals and observations permits the computation of advanced hazard indicators (e.g. damaging potential) that take into account local soil properties. The envelope of synthetic elastic energy spectra reproduces the distribution of the energy demand in the most relevant frequency range for seismic engineering. The synthetic accelerograms can be fruitfully used for design and strengthening of structures, also when innovative techniques, like seismic isolation, are employed. The skill of seismology to estimate realistic ground motions at a particular site should be fully exploited by seismic engineers. In fact, even if recently strong motion records in near-fault, soft soil, or basin conditions have been obtained, their number is still very limited to be statistically significant for seismic engineering applications. The quantification of the critical ground motion expected at a particular site requires the identification of the parameters that characterize the severity and the damage potential. Such critical ground motion can be identified in terms of energy and displacement demands - the latter particularly relevant for seismic isolation, which should be evaluated by considering the seismological, geological, and topographic factors affecting them.