Abstract The extended Canadian Middle Atmosphere Model (extended CMAM) is a general circulation model, which extends from the surface to about 210 km. Spatial complex spectral analysis is applied to horizontal winds simulated by the extended CMAM to obtain semidiurnal tidal amplitudes and phases (from e5 to w5) in the mesosphere and lower thermosphere (MLT) region. The dominant w2 migrating component and the presence of eight nonmigrating tides (w3, w4, w5, e1, e2, e3, e4 and e5) in the mid-latitudes are identified. Components w1 and s0, which tend to maximize at high latitudes, will be discussed separately in a later paper. The migrating semidiurnal tide (w2) has amplitudes reaching over 20 m s −1 for both zonal and meridional winds in the mid-latitude region. Its form compares well to the published results. The amplitudes of nonmigrating semidiurnal tides are non-negligible compared with the migrating semidiurnal tides. The amplitudes for w3 and e2 exceed 12 and 8 m s −1, respectively. Comparisons are made with four nonmigrating semidiurnal components (w3, w4, e1 and e2) derived from the TIMED Doppler interferometer (TIDI) wind measurements between 85 and 105 km altitude and between 45°S and 45°N latitude. Overall, the basic CMAM and TIDI latitudinal structures of the amplitudes agree well and the agreement between the annual mean amplitudes varies with component. Relative to the TIDI results, the CMAM seasonal variations of w4 are in good agreement, of e2 are in reasonable agreement, of w3 are in partial agreement and of e1 are in poor agreement. The 11 semidiurnal components from the model are superimposed to generate the total semidiurnal winds at Jakarta (6°S, 106°E) and Kototabang (0°, 100°E) and are compared with measurements from two equatorial meteor radar stations at these sites. The relative contributions of components to the reconstructed amplitude vary from month to month. The CMAM reconstructions are generally larger than the radar results by a factor varying between one and two. The phases in the radar data are typically stationary with respect to height, whereas they generally decrease with height in the CMAM reconstruction.