We report the first observations of the vertical distribution of hydroxyl (OH) from the upper stratosphere to the mesopause. The Middle Atmosphere High Resolution Spectrograph Investigation (MAHRSI) made these measurements in August 1997. The data confirm the results from the earlier November 1994 MAHRSI mission that were confined to altitudes above 50 km, namely that mesospheric OH densities are 25 to 35% lower than predicted by standard photochemical theory. However, the new observations show that below 50 km the OH density increases rapidly and at 43 km altitude it is larger than that expected from standard theory. This represents a serious dilemma for our understanding of odd-hydrogen chemistry because the same key reactions are thought to dominate OH/HO2 partitioning in both regions. We show that neither standard photochemical theory nor any previously proposed changes are adequate to explain the OH observations in both the upper stratosphere and mesosphere.