The western tropical Pacific is thought to be an important zone for generating El Ni¿o: reflections at the boundary make it a source region of equatorial Kelvin waves. Calculations of the effect of a gappy western boundary on the reflection process are carried out in the framework of the low-frequency limit of the shallow-water equations and thus are highly idealized. The method is also applied to a schematic version of the flow through the Indonesian seas from the western Pacific to the Indian Oceans. The results indicate some strong sensitivities to the location of the gap and to the structure of the incoming flows. In addition, the results can be quite different, depending on whether the zonal extend of the gap is assumed to be infinite or finite. (More precisely, the latter means that the extend of the gap is short compared with the zonal wavelength of the relevant free waves at that frequency.) In view of the complexity of the results for even such a simplified model, it will be very difficult to be confident of any modeling study of the Indonesian throughflow short of a highly resolved numerical calculation with a detailed representation of the geometry and bathymetry. Nonetheless, we offer tentative conclusions concerning the efficiency of the western Pacific boundary as a reflector. Our results suggest that the realistic boundary will not greatly alter expectations based on a simple solid boundary if the reflections important for El Ni¿o are primarily in motion, represented by low-order Rossby modes. This is also consistent with observational evidence indicating no anomalous throughflow during El Ni¿o events. ¿1991 American Geophysical Union