Global magnetospheric simulation results of the field-aligned current (FAC) density at low altitudes are compared with global two-dimensional distributions of Birkeland currents at the topside ionosphere derived from magnetic field data of the Iridium satellite constellation. We present two events with opposite direction of the interplanetary magnetic field By, 23 November 1999 1400--1700 UT, By > 0, and 31 March 2000 1500--1800 UT, By < 0. In both observations and simulations the configuration of the FACs near noon displays the expected reversal of sense with latitude corresponding to the sign of By. The total Region-1 current in the simulations (2--3 MA) is larger than that derived from Iridium (1--1.5 MA), consistent with known underestimates from Iridium, but there is a deficit of Region 2 in the simulations. The poleward limit of the simulated Region-1 currents is significantly, ~5¿, poleward of the observed currents for standard resolution simulations (~6¿ per grid point at 70¿ magnetic latitude). Doubling the simulation resolution leads to a marked improvement in the agreement with the observed spatial distribution of Region-1 FACs. Moreover, the higher resolution leads to the appearance of some Region-2 FACs in the 23 November event. The location of the inner simulation boundary, 2 RE or 3 RE, is found to have negligible effect on the distribution of Region-1 currents, but positioning the inner boundary at 3 RE can be inadequate for capturing Region-2 FACs. Nightside currents in the simulation underestimate the observed FACs. Separating FACs associated with $nabla$ ¿ E, $nabla$ΣP, and $nabla$ΣH terms in the ionosphere, we find that the observed nightside currents agree well with the $nabla$ ¿ E FACs and that the $nabla$ΣH term cancels some of the $nabla$ ¿ E currents on the nightside, resulting in the discrepancy with the observations. The results imply that in addition to including a ring current module to properly account for the particle drift physics, a high simulation resolution and improved nighttime ionospheric conductivities are two factors that also need to be addressed to obtain reliable simulation results.