Implementation of a Raman lidar measurement of middle and upper tropospheric water vapor is described for a system that uses a 532-nm exciting wavelength, fiber-optic signal transfer, and Q-branch selection. Particular attention is given to the minimization of systematic biases introduced by fluorescent reemission of energy associated with elastic backscatter returns. We compare lidar profiles with collocated radiosonde measurements by using the Vaisala H-Humicap capacitive captor. The variations in the water-vapor concentrations on vertical scales of the order of 1 km in the upper troposphere observed by the two instruments present significant differences. Independent characterization of random and systematic lidar measurement errors and radiosonde sensor response characteristics lead to the conclusion that these differences are due to radiosonde sensor response. These intercomparisons indicate that the lidar measurement can provide important information on water-vapor distributions in the radiatively important 8-11-km region.