A three-dimensional objective analysis technique known as CRAM (conditional Relaxation Analysis Method) was applied to investigate various properties of ballistic winds on a mesoscale in mountainous regions. From a 12- day sample of upper-air soundings taken 5 times a day at 2-hr intervals for 12 rawinsonde stations in the Ft. Huachuca region of southeastern Arizona, and artillery firings taken twice a day, CRAM analyses of temperature, density, and winds were performed for 10 atmospheric zones between the surface and 8,000 m using an IBM-7094. It was determined that the CRAM technique produced fields which had the desirable features of map winds, i.e., the contour patterns were relatively smooth and varied slowly with time. The residual deflection errors which resulted were smaller for CRAM (75.2 m) than for a single station (Ft. Huachuca) near the firing range (85.1 m). It was also found that the time decay of ballistic winds in the firing area was smaller using CRAM than using the Ft. Huachuca observation, which implies that CRAM is a better tool with which to make a persistence forecast than a single station.