A large continental normal‐faulting earthquake occurred in the Yinchuan Graben in northern China on 3 January 1739. This event is of significant interest for two reasons. First, it has been suggested on the basis of historical records of shaking that this was a magnitude 8.0 event. If this is true, the 1739 earthquake would be one of the largest continental normal‐faulting earthquakes ever recorded, and this should prompt us to consider the factors contributing to its unusual size. Second, there has been some debate in the literature about the causative structure, with at least three different faults posited as the seismogenic source. We use five¹⁴C ages to date a series of scarps on the East Helanshan Fault, which lies to the west of the graben, and bracket the scarp formation to the last 350 years. We use high‐resolution, stereo imagery from the Pleiades satellites to build a 1 to 2 m resolution DEM of the fault and combine this with field observations to map the East Helanshan rupture in detail. We then construct a throw distribution, determine the slip vector azimuth, and reassess the earthquake magnitude using simple scaling relationships. We conclude that the 1739 earthquake occurred on the East Helanshan Fault, had a rupture length of approximately 87 km, a maximum throw of 5.1 m, and an average throw of 3.0 m, yielding a best estimate magnitude in the rangeM_w 7.1 toM_w 7.6. We suggest that the previous magnitude estimates are biased by enhanced shaking of the sediments within the Yinchuan Graben.

We used airborne light detection and ranging (LiDAR) data to reevaluate the single-event offsets of the 1920 Haiyuan Ms 8.5 earthquake and the cumulative offsets along the western and middle segments of the coseismic surface rupture zone. Our LiDAR data indicate that the offset observations along both the western and middle segments fall into groups. The group with the minimum slip amount is associated with the 1920 Haiyuan Ms 8.5 earthquake, which ruptured both the western and middle segments. Our research highlights two new interpretations: First, the previously reported maximum displacement of the 1920 earthquake was likely due to at least two earthquakes; second, our results reveal that the cumulative offset probability density (COPD) peaks of the same offset amounts on the western and middle segments do not correspond to one another one-to-one. We suggest that any discussion of the rupture pattern of a certain fault based on the offset data should also consider fault segmentation and paleoseismological data. Therefore, the COPD peaks should be computed and analyzed on fault subsections and not entire fault zones to study the number of paleoearthquakes and their rupture patterns.