A climatology of the eddy dissipation rate is presented based on the widths of the Doppler radar spectra observed by the 50-MHz, clear-air profiler located at White Sands Missile Range, New Mexico. The observations are from radar beams 15¿ off zenith and extend from about 5 to 20 km altitude. The data are nearly continuous over the 5-year period 1991--1995. Mean corrections to the observed spectral widths are much larger for beam-broadening effects than for vertical shear of the wind or gravity-wave effects. The mean eddy dissipation rate falls between about 10-3.5 and 10-2.5 m2 s-3 at all altitudes and all seasons. The hourly means at a given height have an approximately lognormal frequency distribution. The hourly values have larger standard deviations in the troposphere than in the stratosphere. The maximum eddy dissipation rate is found at the lowest altitude observed, and the minimum value is found about 2--3 km below the tropopause, depending upon season. At about 6--10 km altitude the summer values of eddy dissipation rate are several decibels larger than those during other seasons. When the observations are sorted according to distance from the tropopause, there is very little difference among the seasonal values. The summer tropospheric values show very large interannual changes; the large values of eddy dissipation rate correlate with high surface temperatures and with large mean hourly standard deviations of vertical velocity (an indicator of convective activity in the troposphere). The diurnal changes of eddy dissipation rate and of hourly vertical velocity standard deviations also have similar patterns; that is, largest values occur in the afternoon, and smallest values occur during the night. The largest diurnal range of eddy dissipation rate is found during the summer in the troposphere. The correlation of hourly values of eddy dissipation rate with hourly values of wind speed and vertical wind shear is relatively low, which suggests that the eddy dissipation rate is strongly controlled by the small-scale flow and is not well predicted from features of the large-scale flow.¿ 1997 American Geophysical Union