It is shown that although the power spectra of the AE and PC data are quite similar, they appear different in the structure function (SF) analysis. While the AE time series has a clear drop in the slope of the SF after the first 2 hours, the slope of the SF of the PC data only decreases gradually, and at a somewhat longer timescale. It is also shown by using 15-min averaged data, that both SFs are periodic with a clear diurnal variation. The PC time series seems to have a more pronounced periodicity, probably because it is measured at a single station at Thule. The relation between the PC and AE indices has been analyzed in detail. The cross-correlation time between these data is shown to be 2--2.5 hours. It appears that the correlation function has two different scaling regions, one at the timescale within 2.5 hours and another between 2.5 and 12 hours. It also seems that a neural network prediction of the AE index from the PC index is only possible for time scales shorter than the cross-correlation time of these systems. The AE index has been derived from the PC index for 7.5 min ahead. These predictions gave normalized mean square errors (NMSE) in the range 0.18--0.25 during wintertime. The corresponding correlation coefficient was 0.91 at the best. This shows that well within the cross-correlation time, the AE time series can be derived quite accurately from the PC data, at least during wintertime when the field-aligned currents are the main source of the PC index. ¿ 1998 American Geophysical Union