Geomagnetic and auroral activity vary seasonally with maxima at equinoxes, as has been known for more than a century. The cause remains under debate. The angle made by the Earth's dipole axis with the typical direction of the interplanetary magnetic field (IMF) can explain a portion (about 17%) of the effect. To explain the majority of the equinoctial effect, we suggest that geomagnetic activity peaks when the nightside auroral zones of both hemispheres are in darkness, as happens at equinox. Under such conditions, no conducting path exists in the ionosphere to complete the currents required by solar wind-magnetosphere-ionosphere coupling, and geomagnetic disturbances maximize. To test this theory, the Universal Time (UT) variation of geomagnetic activity was explored. As our model predicts, geomagnetic activity in December, measured by the Am index, evinces a deep minimum around 0300-0600 UT when the auroral oval of both hemispheres are in darkness and a maximum around 1500--1600 UT when the southern nightside oval is sunlit. In June, complementary effects are predicted and observed. Previous studies using the AE index have shown more ambiguous results. Here we show that if AE is resolved into the AU and AL components, the discrepancy disappears, with the AL component following the same pattern as does Am. We thus conclude that the intensity of global geomagnetic activity is well ordered by whether the nightside auroral oval is sunlit in one hemisphere or neither. ¿ 2001 American Geophysical Union