The 9.5-year temperature time series from the Halogen Occultation Experiment (HALOE) aboard the Upper Atmosphere Research Satellite (UARS) has been analyzed for 40¿N, 30¿N, 20¿N, Tropics, 20¿S, 30¿S, and 40¿S using multiple linear regression techniques at 12 pressure levels from 5 to 0.01 hPa. The HALOE sunset and sunrise temperature profiles have good vertical resolution and are representative of the variations of the zonal average state. Vertical profiles with latitude of the average sunset/sunrise differences agree qualitatively with that expected from the effects of tides. Periodic seasonal (annual and semiannual) terms are sampled well in the time series, and the seasonal cycle from HALOE at 40¿N agrees well with the local variations reported from measurements with the ground-based Rayleigh lidar instrument at 44¿N, 6¿E. Significant quasi-biennial oscillation (QBO)-like variations occur at most latitudes of the upper stratosphere in the HALOE data, and their phases are nearly symmetric about the equator. There is also a clear subbiennial (688-day) cycle at Northern Hemisphere (NH) midlatitudes at those same pressure altitudes due to the interaction of the QBO and annual cycles. The solar cycle temperature term is in-phase with the UV flux forcing in the upper stratosphere but lags it by a year or so in the upper mesosphere at NH midlatitudes. There is also a significant cooling trend for the HALOE time series that is of order 1.0 to 1.6 K per decade in the midmesosphere at 20¿N and at 1 hPa in the tropics. The seasonal terms may be combined with the constant and autoregressive (AR) terms (or annual average distribution that we provided) in order to generate the monthly zonal average HALOE climatology of middle atmosphere temperature for the decade of the 1990s. Comparisons between the HALOE and CIRA climatologies for the upper stratosphere suggest a cooling of order 2.5 K per decade from the mid-1970s to the mid-1990s. HALOE/CIRA differences for the mesosphere change sign with altitude, an indication of bias errors in one or both.