Abstract. We have analysed a database of ∼300 h of tristatic ionospheric velocity measurements obtained overhead at Tromsø (66.3° magnetic latitude) by the EISCAT UHF radar system, for the presence of flow effects associated with the y-component of the IMF. Since it is already known that the flow depends upon IMF B_z, a least-squares multivariate analysis has been used to determine the flow dependence on both IMF B_y and B_z simultaneously. It is found that significant flow variations with IMF B_y occur, predominantly in the midnight sector (∼2100–0300 MLT), but also pre-dusk (∼1600–1700 MLT), which are directed eastward for IMF B_y positive and westward for IMF B_y negative. The flows are of magnitude 20–30 m s^–1 nT^–1 in the midnight sector, and smaller, 10–20 m s^–1 nT^–1, pre-dusk, and are thus associated with significant changes of flow of order a few hundred m s^–1 over the usual range of IMF B_y of about ±5 nT. At other local times the IMF B_y-related perturbation flows are much smaller, less than ∼5 m s^–1 nT^–1, and consistent with zero within the uncertainty estimates. We have investigated whether these IMF B_y-dependent flows can be accounted for quantitatively by a theoretical model in which the equatorial flow in the inner magnetosphere is independent of IMF B_y, but where distortions of the magnetospheric magnetic field associated with a "penetrating" component of the IMF B_y field changes the mapping of the field to the ionosphere, and hence the ionospheric flow. We find that the principal flow perturbation produced by this effect is an east-west flow whose sense is determined by the north-south component of the unperturbed flow. Perturbations in the north-south flow are typically smaller by more than an order of magnitude, and generally negligible in terms of observations. Using equatorial flows which are determined from EISCAT data for zero IMF B_y, to which the corotation flow has been added, the theory predicts the presence of zonal perturbation flows which are generally directed eastward in the Northern Hemisphere for IMF B_y positive and westward for IMF B_y negative at all local times. However, although the day and night effects are therefore similar in principle, the model perturbation flows are much larger on the nightside than on the dayside, as observed, due to the day-night asymmetry in the unperturbed magnetospheric magnetic field. Overall, the model results are found to account well for the observed IMF B_y-related flow perturbations in the midnight sector, in terms of the sense and direction of the flow, the local time of their occurrence, as well as the magnitude of the flows (provided the magnetic model employed is not too distorted from dipolar form). At other local times the model predicts much smaller IMF B_y-related flow perturbations, and thus does not account for the effects observed in the pre-dusk sector.

Key words: Magnetospheric physics (magnetosphere · ionosphere interactions) – Ionosphere (plasma convection; auroral ionosphere)