Some interplanetary studies lead one to expect that the toroidal fields of individual active regions are directly related to their heliospheric structure. Other studies conclude that the large-scale solar dipolar field is more important. We have carried out two studies that bear on these apparently conflicting views. We first studied geomagnetic events temporally associated with the eruption of 18 individual coronal X-ray sigmoids, which occurred while the large-scale solar dipolar magnetic field pointed southward. We found that if a coronal flux rope model is used to interpret magnetic structure, eruptions with a southward leading magnetic field are associated with stronger geomagnetic storms, but those with a northward leading field are associated with more storms. We next studied three full magnetic cycles, solar cycles 17--22. We examined the temporal variation of the ratio of the geomagnetic Ap index to the sunspot number. We found no statistically compelling fluctuations of this quantity on solar cycle time scales that are in phase with the reversal of active region polarities. On the other hand, we found a weak tendency for fluctuations that are in phase with the reversal of the large-scale solar dipole field. From these two studies we infer that the magnetic structure of individual active regions plays a role in geomagnetic events, but their geoeffectiveness is complicated by asymmetries in the leading and following magnetic field and density. We conclude that simple cycle-dependent generalizations have only statistical significance, and cannot dependably be used to predict the geomagnetic effects of a given solar eruption. ¿ 2001 American Geophysical Union