The water vapor differential absorption lidar (DIAL) of the German Aerospace Research Establishment (DLR) was flown aboard the National Center for Atmospheric Research (NCAR) Electra research aircraft during the Boreal Ecosystem-Atmosphere Study (BOREAS). The downward looking lidar system measured two-dimensional fields of aerosol backscatter and water vapor mixing ratio in the convective boundary layer (CBL) and across the CBL top (zi). We show a case study of DIAL observations of vertical profiles of mean water vapor, water vapor variance, skewness, and integral scale in the CBL. In the entrainment zone (EZ) and down to about 0.3 zi the DIAL observations agree with in situ observations and mixed-layer similarity theory. Below, the water vapor optical depth becomes large and the DIAL signal-to-noise ratio degrades. Knowing the water vapor surface flux and the convective velocity scale w* from in situ aircraft measurements, we derive entrainment fluxes by applying the mixed-layer gradient (MLG) and mixed-layer variance (MLV) methods to DIAL mixing ratio gradient and variance profiles. Entrainment flux estimates are sensitive to our estimate of zi. They are shown to be rather insensitive to the input surface flux and to the DIAL data spatial resolution within the investigated range. The estimates break down above about 0.9 zi as the flux-gradient and flux-variance relationships were developed to describe the large-scale mixing in the mid-CBL. The agreement with in situ entrainment flux estimations is within 30% for the MLV method. On a flight leg with significant mesoscale variability the entrainment flux turns out to be 70% higher than the in situ value. This is in good agreement with the fact that large-eddy simulations (LES) of mean water vapor profiles and variances, upon which the MLG and MLV methods are based, do not include mesoscale variability. The additional water vapor variance from mesoscales may then lead to the overestimate of the flux. Deviations from the in situ observations may also be due to poor LES resolution of small-scale mixing in the EZ, similarly coarse resolution of the DIAL data, or a capping inversion in the LES model (8 K) which is significantly stronger than the observed inversion (3--4 K). ¿ 1997 American Geophysical Union