A novel approach to measure the wind flow field in a utility-scale wind farm is described. The measurement technique uses a mobile, three-dimensional scanning LiDAR system to make successive measurements of the line-of-sight (LOS) wind speed from three different positions; from these measurements, the time-averaged three-dimensional wind velocity vectors are reconstructed. The scanning LiDAR system is installed in a custom-built vehicle in order to enable measurements of the three-dimensional wind flow field over a footprint that is larger than with a stationary scanning LiDAR system. At a given location, multiple series of plan position indicator (PPI) and velocity azimuthal display scans are made to average out turbulent fluctuations; this series is repeated at different locations across the wind farm. The limited duration of the total measurement time period yields measurements of the three-dimensional wind flow field that are unaffected by diurnal events. The approach of this novel measurement technique is first validated by comparisons to a meteorological mast and SODAR at a meteorological observatory. Then, the measurement technique is used to characterize the wake flows in two utility-scale wind farms: one in complex terrain and the other in flat terrain. The three-dimensional characteristics of the wakes are described in the measurements, and it is observed that in complex terrain the wake has a shorter downstream extent than in flat terrain. A maximum deficit in the wind speed of 20–25% is observed in the wake. The location of the maximum deficit migrates upward as the wake evolves; this upward migration is associated with an upward pitching of the wake flow. A comparison of the measurements to a semi-empirical wake model illustrates how the measurements, at full-scale Reynolds numbers, can support further development of wake models.