A combined spatial and secular variation model of the geomagnetic field, labeled M061581, is derived from a selection of Magsat data. Secular variation (SV) data computed from linear fits to midnight hourly values from 19 magnetic observatories were also included in the analysis but were seen to have little effect on the model. The SV patterns from this new model are compared with those from the 1980 IGRF and with those for 1970 computed by the DGRF and with the 1960 patterns computed using the GSFC(12/66) model. Most of the features of the M061581 are identical in location and level with those of the 1980 IGRF. Together they confirm that the reversals in sign of field change seen over Asia and North America between 1965 and 1975 are reverting to the pre-1965 states. The M061581 model gives −32 nT/yr for the dipole decay rate, larger than the 70% increase already reported since 1965. This abnormally high value is interpreted as being a defect of the model because it appears to result from a much larger (−100 nT/yr) drop in field over the polar regions not indicated by the 1980 IGRF. This north polar decrease is shown to be of external origin as the result of a combination of the seasonal effect of the north polar ionospheric (counterclockwise) afternoon Sq^p cell increasing in intensity from the beginning (November 1979) to the end (June 1980) of the Magsat data collection period, coupled with an enhancement of its effect as the orbit lowered from the 350– 550 km initial altitudes to near 200 km just prior to burnup. This experiment indicates that secular variation can be obtained from satellite data for intervals of less than a full year if corrections can be made for seasonal effects and that ‘annual snapshots’ of the field by a satellite would allow easy and accurate models of secular change without the use of any surface data.