The encounter of a deep western boundary current with the equator is studied using an ultra-high resolution numerical model of the equatorial ocean in a study of the origin of the deep equatorial current system. The integration was carried out from a state of rest to 3500 days. The spin-up experiment shows that the deep western boundary current, upon encountering the equator, first turns along the equator. The time scale of spin-up is fast along the equator and the boundaries (about 250 days) but is considerably slower off the equator. After this initial response, the western boundary current slowly begins to penetrate into the opposite hemisphere, as expected from the linear theory. The eastward equatorial current gradually slows down: the total water mass transport in the equatorial branch decreases steadily until by day 3000 it has dwindled to zero and all the mass transport in the western boundary current crosses into the other hemisphere. However, a vestige of the equatorial branch remains until the end of the integration with the speed of less than 3 cm/s and with intermittent meandering. An analysis of the momentum balance indicates that this meander may be a free equatorial inertial oscillation. We postulate that a permanent, steady bifurcation of the mid-depth western boundary current that carries substantial mass does not occur and eventually all of the mass transport will continue into the opposite hemisphere. The asymptotic flow field seems incompatible with the tracer field observed in the Atlantic, which clearly shows a splitting at the equator of the trace signal associated with the Upper North Atlantic Deep Water. We speculate that an inherently time-dependent explanation of the tracer tongue is called for. ¿ American Geophysical Union 1992