The initial goal of the three-dimensional (3-D) vertical seismic profiling (VSP) at Lawrence Livermore National Laboratory (LLNL) was to characterize seismic wave velocities and frequencies below the vadose zone in order to design the acquisition geometry for a high-resolution 3-D seismic reflection survey. VSPs are also used routinely to link surface seismic data with well logs. However, a test of the two-dimensional (2-D) seismic line recorded at the LLNL Livermore Site in the spring of 1994 indicated that obtaining high-quality reflection images below the vadose zone, but shallower than about 160 ft, would require an expensive, very finely sampled survey ({gt} 1-m receiver spacing). This paper presents the difficulties encountered during initial data acquisition and processing, and attempts to alleviate the difficulties in the field and laboratory.