Candela et al. <2003> have reported empirical orthogonal function (EOF) analyses based on 23-month current-meter and acoustic Doppler current profiler measurements in the Yucatan Channel. Those authors noted the difference between EOFs obtained from observations and their z-level models and EOFs calculated by Ezer et al. <2003> from the results of a terrain-following model. Here a new analysis is reported that explains this difference, and that also suggests the importance of shelf-edge meander mode of the core Loop Current in the channel. We show that the terrain-following model gives EOFs with characteristics similar to those observed when data from the upper slope and shelf in the western portion of the model channel are omitted. Modes 1 and 2 have tripole and dipole structures with energies (35%, 26%), respectively, of total energy, and correlate with slow vacillation of the core-current for periods >50 days. Exclusion of upper-slope and shelf data eliminates a short-period and energetic component inherent in Ezer et al.'s original mode 1 EOF. This mode correlates with frontal meanders of the core current over the shelf edge in the western portion of the channel. The short-period mode may be missing or underestimated in observational and z-level models' analyses, since there were only a few moorings over the upper slope and shelf, and z-level models have step-like topography with generally lower resolution in shallower seas.