Based on observations at a polar site (69°N), Lübken and von Zahn (1991) reported a counter-intuitive bistable pattern of mesopause, high in winter (192K at 98 km) and low in summer (129K at 88 km). Though unexpected, Na lidar observations showed that the bistable pattern also exists in a midlatitude station (40.6°N) with winter mesopause at 101 km and summer mesopause at 86 km (She et al., 1993;. The two-level mesopause was further confirmed by shipborne K lidar observations (von Zahn & Höffner, 1996), which enabled observations between 71°S and 54°N from late April to early July 1996. This set of measurements which included southern and northern equatorial regions showed that the summer state does not exist in the equatorial region. Using 2 years of data (1996)(1997) from Fort Collins (40.6°N, 105°W) and Kühlungsborn, Germany (54°N, 12°E), along with older monthly mean data from ALOMAR, Norway (69°N, 16°E), the concept of the two-level mesopause was further elucidated (She & von Zahn, 1998) and idealized in Figure 1 of their paper. They asserted the two-level mesopause structure exists globally in midlatitude and polar regions but not in equatorial regions.In the two-level structure, there exists an altitude of minimum annual variation in temperature near 98 km and an altitude of maximum annual variation in temperature near 86 km. There also exists a high-altitude winter mesopause near 100 km and a low-altitude summer mesopause near 88 ± 3 km with a clear and abrupt jump between them, despite of the fact that unlike at 69°N and 54°N, the sharp summer/winter transitions were not observed at 40.6°N. The less clear summer/winter mesopause transition at 40.6°N is due to the relatively smaller difference between summer and winter mesopause temperatures at a lower-latitude station, which may be over-powered by superimposed day-to-day wave perturbations in a 2-year data set. However, when the true climatology of the temperature structure is revealed, the proposed two-level mesopause structure with sharp winter/summer transitions can be ascertained. One objective of this paper is to use the 20 years (956 nights) of nocturnal temperature observations to investigate the climatology of the two-level mesopause and to reveal the sharp mesopause transitions at midlatitude.