The calculations for seismic hazard can be quite complex. First, the regional geology and seismology is examined for patterns. Zones of similar potential for seismicity are drawn. For example, the famous San Andreas Fault might be drawn as a long narrow zone. Zones in the continental interior (the site for intraplate earthquakes) would be drawn as broad areas, since causative faults are generally not identified.
Each zone is given properties associated with source potential: how many earthquakes per year, the maximum size of earthquakes, etc. Finally, the calculations require formulae that give the required hazard indicators for a given earthquake size and distance. For example, some districts prefer to use peak acceleration, others use peak velocity, and more sophisticated uses require response spectral ordinates.
The computer program then integrates over all the zones and produces probability curves for the key ground motion parameter. The final result gives you a 'chance' of exceeding a given value over a specified amount of time. Standard building codes for homeowners might be concerned with a 1 in 500 years chance, while nuclear plants look at the 10,000 year time frame.
More elaborate variations on the theme also look at the soil conditions. If you build on a soft swamp, you are likely to experience many times the ground motions than your neighbour on solid rock.
The US seismic hazard maps http://geohazards.cr.usgs.gov/eq/