The skin friction and cross-flow separation location on a non-body-of-revolution (non-BOR) ellipsoidal model performing constant-yaw turns and a pitch-up maneuver, each with roll oscillation were studied for the first time. The detailed, low uncertainty, flow topology data provide an extensive experimental database on the flow over non-BOR hull shapes that does not exist anywhere else in the world and serves as a crucial tool for computational validation. The ellipsoidal model was mounted on a roll oscillation machine in the Virginia Tech Stability Wind Tunnel slotted wall test section. Hot-film sensors with constant temperature anemometers provided skin friction magnitudes on the body's surface for thirty-three steady flow model orientations and three unsteady maneuvers at a constant Reynolds number of 2.5 million. Cross-flow separation locations on the model were determined from span-wise minima in the skin friction magnitude for both the steady orientations and unsteady maneuvers. Steady hot-film data were obtained over roll angles between ±25° in 5° increments with the model mounted at 10° and 15° yaw and at 7° pitch with respect to the flow. The roll oscillation machine was used to create a near sinusoidal unsteady roll motion between ±26° at a rate of 3 Hz, which corresponded to a non-dimensional roll period of 5.4. Unsteady data were obtained with the ellipsoidal model mounted at 10° and 15° yaw and at 7° pitch during the rolling maneuver. Cross-flow separation was found to dominate the leeside flow of the model for all orientations. For the yaw cases, the separation location moved progressively more windward and inboard as the flow traveled downstream. Increasing the model roll or yaw angle increased the adverse pressure gradient on the leeward side, creating stronger cross-flow separation that began further upstream and migrated further windward on the model surface. For the pitch flow case, the cross-flow separation remained straight as the flow moved axially downstream. The strongest pitch cross-flow separation was observed at the most negative roll angle and dissipated, moving further downstream and inboard as the model's roll angle was increased. The unsteady flow maneuvers exhibited the same flow topology observed in the quasi-steady conditions. However, the unsteady skin friction and separation locations lagged their quasi-steady counterparts at equivalent roll angles during the oscillation cycle. A first order time lag model and sinusoidal fit to the separation location data quantified the time lags that were observed. / Ph. D.