Magnetosheath plasma penetrates into the magnetosphere creating the particle cusp, and similarly the interplanetary magnetic field (IMF) By component penetrates the magnetopause. We reexamine the phenomenology of such penetration to investigate implications for the magnetopause merging site. Three models are popular: (1) the ''antiparallel'' model, in which merging occurs where the local magnetic shear is largest (usually high magnetic latitudes); (2) a tilted merging line passing through the subsolar point but extending to very high latitudes; or (3) a tilted merging line passing through the subsolar point in which most merging occurs within a few Earth radii of the equatorial plane and local noon (subsolar merging). It is difficult to distinguish between the first two models, but the third implies some very different predictions. We show that properties of the particle cusp imply that plasma injection into the magnetosphere occurs most often at high magnetic latitudes. In particular, we note the following: (1) The altitude of the merging site inferred from midaltitude cusp ion pitch angle dispersion is typically 8--12 RE. (2) The highest ion energy observable when moving poleward through the cusp drops long before the bulk of the cusp plasma is reached, implying that ions are swimming upstream against the sheath flow shortly after merging. (3) Low-energy ions are less able to enter the winter cusp than the summer cusp. (4) The local time behavior of the cusp as a function of By and Bz corroborates predictions of the high-latitude merging models. We also reconsider the penetration of the IMF By component onto closed dayside field lines. Our approach, in which closed field lines move to fill in flux voids created by asymmetric magnetopause flux erosion, shows that strict subsolar merging cannot account for the observations. ¿ American Geophysical Union 1995