Using more than 145 hours of nighttime lidar measurements obtained during October 3--16, 1996, and October 2--11, 1997, the tidal signature in the middle atmospheric thermal structure (15--95 km) at Mauna Loa, Hawaii (19.5 ¿N), was investigated. The daytime High Resolution Doppler Imager (HRDI) temperatures taken in September and October 1993--1997 and zonally averaged at the same latitude were also used. The daytime HRDI and nighttime lidar temperature differences from their respective daytime and nighttime averages were compared to the equivalent differences predicted by the Global Scale Wave Model (GSWM) at the same latitude. Some consistent local solar time (LST)-related structures were observed in both HRDI and lidar data, suggesting the presence of important migrating tidal components. In particular, a warm period was clearly identified, propagating downward from 105 km at 0800 LST to 65 km at 0000 LST and surrounded by two colder periods above and below. These warm/cold periods were predicted to occur 2 to 3 hours later by GSWM compared to the HRDI observations. Other LST-related structures were observed by lidar between 30- and 80-km altitude, in particular, a colder early night, warmer midnight, and colder late night around ~70 km, suggesting a significant semidiurnal component at this altitude. As previously observed, the amplitudes predicted by GSWM were much smaller than those observed by lidar and HRDI. A new analysis constrained wave adjustment method described in a companion paper <Leblanc et al., this issue> was used to estimate the diurnal and semidiurnal components from the nighttime-only lidar data. The main point of disagreement between the lidar observations and GSWM predictions occurred between 60 and 85 km. A large semidiurnal component was observed by lidar, leading to early and late cold night and warm midnight, while no such large semidiurnal component was predicted by GSWM, leading to an apparent warm early night at 60 km and an apparent cold midnight at 80 km and above. It appears that the tidal structure observed by lidar is more representative of that predicted by GSWM at 24 ¿N, suggesting a latitudinal shift between theory and observation. It is not clear whether this shift is related to an indetermination of the tidal source and/or propagation or if the observed differences are simply due to oscillations related to local/regional local solar time obscuring the global tidal signature. ¿ 1999 American Geophysical Union