AbstractPaleoseismic rupture histories provide spatiotemporal models of earthquake moment release needed to test numerical models and lengthen the instrumental catalog. We develop a model of the fewest and thus largest magnitude earthquakes permitted by paleoseismic data for the last 1,500 years on the southern San Andreas and San Jacinto Faults, California, USA. The largest geometric complexity appears to regulate the system: Only two ruptures break the San Gorgonio Pass region, followed by episodes of ruptures that could bridge the northern San Jacinto Fault and the San Andreas Fault. When tested against independent data on slip per event, the model produces comparable values indicating the end‐member model does not underpredict rupture rates. Rupture of >85% of the fault length in the historic period between 1800 and 1857 and the subsequent quiescence is similar to epochs of activity in the prehistoric model, suggesting that regional clustering of seismicity could be a trait of the system.