This research evaluates models of earthquake behavior and fault segmentation, by quantifying the rupture-magnitude pattern along the trace of coseismic surface rupture for each of four earthquakes. One is the 1983 (Ms = 7.3) Borah Peak, Idaho, earthquake and three are prehistoric. Two earthquakes occurred along each of the adjacent Thousand Springs and Mackay Segments (TSS and MS) of the Lost River fault in east-central Idaho, and within their intervening segment boundary. The apparent vertical slip-component (throw) was measured as the vertical separation of surveyed longitudinal profiles of faulted stream floodplains, terraces, and alluvial fans, and relative uncertainty in the measurements was obtained. The true vertical slip-component (VD) was calculated to account for geometric distortion in throw data. Knowledge of fault dip is required, and was estimated using structural contour models of the fault. The preferred model relies on the measured orientation of the 1983 net-slip vector at a faulted-fence site, but the calculation is more dependent on profile gradient then on fault dip. The ages of faulted landforms are constrained by a dated soil chronosequence (in calibrated calendar years BP). The period of record is approximately ≈ 17 ky, the age of ubiquitous late-glacial fans. The Borah Peak surface rupture is 5 km longer than previously thought. 1983 rupture VD was uniformly approximately ≈ 2 m along the southeastern half of the TSS and penetrated the segment boundary with uniform VD of 0.5 m. The penultimate earthquake on the TSS occurred between 10 and 11 ka, and had rupture magnitude and location nearly identical to the 1983 earthquake. On the Mackay Segment (MS), an earthquake occurred at approximately ≈ 5 ka with ruptures penetrating the segment boundary. Apparently a previous event occurred between 12 and 17 ka. The cumulative VD for the period of record is uniform 4 to 4.5 m on both the TSS and the MS, but only 2 m ofVD in the segment boundary. There is no evidence for medium sized earthquakes. All the evidence can be explained by four nearly identical, large magnitude characteristic earthquakes on a segmented fault.