Geodetic survey measurements are used to estimate the coseismic slip distribution in the 1944 Tonankai (Mw=8.1) and 1946 Nankaido (Mw=8.3) earthquakes and to assess quantitatively the degree to which this slip is resolved on the plate boundary megathrust. Data used include 798 angle changes from triangulation surveys, 328 leveling section differences, and 5 coseismic tidal gage offsets. Many of the nominally coseismic triangulation data span ~50 years, nearly half the earthquake cycle, and correction for interseismic deformation using post-1950 observations is applied. Microseismicity is used to define the configuration of the plate boundary interface and approximate it with a continuous, multisegment fault model. Because the onshore geodetic data have very limited resolving power for offshore fault segments, offshore coseismic slip was constrained by Satake's <1993> estimation based on tsunami data. The majority of the coseismic slip occurs between 15 and 25 km depth. Although resolution declines toward the trench axis, it is sufficiently good to define two distinct high-slip regions, one off southeastern Shikoku Island (11 m maximum) and the other offshore of Kii Peninsula (3 m maximum). The slip magnitude off southeastern Shikoku, coupled with the plate convergence rate, would imply an recurrence interval of about 270 years, much longer than the average repeat time of ~120 years for historical great earthquakes on the Nankai Trough. However, the maximum coseismic slip is sensitive to the assumed fault geometry, and slippage on trough-parallel splay faults could significantly decrease the maximum slip to about 6 m. ¿ 1999 American Geophysical Union