The nature of the targets responsible for certain clear-air dot angel echoes, and their suitability as wind tracers, are deduced from pulse Doppler radar observations of their velocity characteristics. The angel echoes were observed during a six-hour period in the lowest few thousand feet of the atmosphere. They were all discrete point targets and probably had discrete Doppler velocities as well (i.e., they were coherent). The angel populations showed well-defined swarm velocities with superimposed random deviations. Three types of dot angels were distinguished according to their mean deviation from the swarm velocity and their average vertical motion. Type 1, in the early afternoon, showed mean velocity deviations of 1–15 m sec−1 and average downward motions of up to 4.5 m sec−1. Type 2, in the late afternoon, showed mean deviations of less than 1 m sec−1 and average downward motions of less than 1.8 m sec−1. Type 3, after sunset, showed mean deviations of 1–4 m sec−1 and average vertical motions that were mainly upward at up to 3.7 m sec−1. The small back-scattering cross sections of individual angels (≲ 10−1 CM2), their discreteness in space and velocity, their often quite large mean deviations from a uniform velocity, and the fact that the only major upward velocities occurred after sunset, at a time when the lapse rate was becoming increasingly stable, all suggest insects rather than atmospheric inhomogeneities as the source of the angels. The Type 1 angels had mean swarm velocities differing from the wind by less than 2 m sec−1; Type 2 probably differed by even less than this and are judged to have been good tracers of the wind. The Type 3 angels, on the other hand, had swarm velocities of up to 5 m sec−1 relative to Type 2 and hence to the wind as well, and thus they were poor tracers of the wind.