Under favorable conditions, georadar techniques can provide vivid images of the shallow subsurface (<10-50 m). Although the significant advantages of 3-D georadar surveying strategies are well documented, they generally require much greater expenditures than traditional 2-D approaches. We introduce an efficient, semi-automated 3-D georadar acquisition and processing scheme that does not jeopardize data quality. A standard georadar acquisition unit is integrated with an innovative self-tracking laser theodolite with automatic target recognition capabilities. Georadar and coordinate data are recorded simultaneously as the georadar antennae are transported steadily across a survey area. While tracking a target prism mounted between the antennae, the theodolite provides real-time coordinate information with high accuracy in all directions. At approximately 1 m/s (moderate walking speed) the coordinates are determined to better than + or -0.04 m, and at approximately 3 m/s they are better than + or -0.07 m. Subsequent to acquisition, semi-automatic processing allows the georadar data to be static corrected, transferred to a regular grid using a novel 2-D Fourier transform method, amplitude modulated, filtered, and displayed. The acquisition component of the new scheme is five to ten times faster than standard step-mode georadar techniques, and the semi-automatic processing component allows initial 3-D images to be viewed in the field. Typically, a 3-D georadar data set covering a 25 mX25 m area may be collected and processed in less than 3 hours. One such data set recorded across a former glaciofluvial environment allows reflections from complex river channel sediments and surface features to be readily identified and interpreted.