A series of in-situ temperature measurements performed by falling spheres at high latitudes in the last ~10 years is compared with historical temperature data collected by the rocket grenade technique in the mid 1960s. Concentrating on summer (here from mid May to mid August) and on altitudes between 50 and 85 km, where natural variability and instrumental uncertainties are small, the observed trend is very small, if present at all. Taking into account all data from 50 to 85 km the mean trend is -0.024 K/y with a statistical error of ¿0.014 K/y, i.e., practically zero. The mean deviation of the grenade temperatures from the falling sphere data is 0.8¿6.4 K and the differences are randomly distributed. We have also analyzed various subsets of the entire data set, e.g., in a restricted altitude range where lidars have shown large trends at mid latitudes, namely from 55 to 75 km. The main result remains unchanged: The deviation between 'old' and 'new' data is practically zero and the temperature trend in the high latitude summer mesosphere is certainly much smaller (if any at all) compared to that observed at mid latitudes. ¿ 2000 American Geophysical Union