A proposed ballistic wind determination technique using the satellite infrared spectrophotometer (SIRS) data from NIMBUS 3 and 4 was tested. The general approach was to derive regression equations from a dependent data sample of nearly coincident subsatellite points and rawinsonde locations. The validity of the equations was checked with a smaller independent sample of SIRS radiances from a later period. Tests with the ballistic wind component normal to the path of the NIMBUS 3 suggested standard errors of less than 10 kts can be achieved. Both zonal and meridional components can be calculated from SIRS-B radiance data which is grouped in clusters. Because the geostrophic ballistic wind is proportional to a horizontal gradient, several techniques were attempted for calculating the gradient at the central point of the irregularly spaced set of data. (Author)