The Ice, Cloud and land Elevation Satellite (ICESat) provides the first opportunity for measurement of surface elevation hi over the portions of the Antarctic ice shelves that are south of the European Remote Sensing (ERS) satellite maximum latitude (81.5¿S). The dominant source of short-period variability in hi is ocean tides. We use crossover elevation difference (Δhi) data from the Ross Ice Shelf (RIS) to demonstrate ICESat's ability to detect the tidal signal, and to compare the accuracy of several tide models. The root-mean-square (rms) value of all RIS measurements of Δhi is ≈0.74 m; after removing the tide using the most accurate model, the rms of the residual signal in regions of optimal model performance is ≈0.16 ¿ 0.03 m. This value corresponds to an uncertainty in hi of 0.11 ¿ 0.02 m. We postulate that the primary sources of the residual signal are tide model errors and the inverse barometer effect.