We present a new method to objectively combine paleoseismic event data from multiple sites into rupture scenarios and apply it to the southern San Andreas fault (SSAF) of California. First, a pool is constructed of all ruptures between sites allowed by fault geometry and available event age probability distribution functions (PDFs). Scenarios constructed by drawing from this pool are evaluated by the average quality of agreement event dating evidence, by the degree of misfit of cumulative displacement over all ruptures compared to a prediction from the fault slip rate and elapsed time, and by the number of events in the scenario. Three slip-rate models from the 2008 Working Group on California Earthquake Probabilities (WGCEP) were considered. Scenarios with full fault length ruptures tend to be inconsistent with low slip rates through the San Bernardino and San Gorgonio sections of the SSAF. Favorable scenarios tend to include 1857-like ruptures for three of the most recent five prehistoric ruptures in the northern half of the SSAF from Carrizo to approximately Wrightwood. They also include one to two ruptures that involve the southern half of the SSAF, but most earthquakes there appear shorter and exhibit less consistency from event to event. By combining paleoseismic data into ensembles of scenarios and selecting viable scenarios using external constraints, our method provides rupture histories useful for seismic hazard assessment without having to first settle which event at a site correlates with those at adjoining sites. This opens a way for paleoseismic data to be used with greater power to understand the seismic hazard posed by faults like the southern San Andreas.