A scale model experiment was conducted to study the effect of topography on ground motion for incident P waves. It was found that amplifications as large as 35 per cent may occur in the time domain for a wide-band input with predominant wavelength equal to the width of the scatterer base. The largest diffracted wave observed was a forward scattered Rayleigh wave with an amplitude of about 55 per cent of the incident motion. Examination of the spectral ratios showed that, for more band-limited input pulses, peak amplifications of the crest motions relative to the base may be as high as 200 per cent. These differences are comparable to, but smaller than, the 400 per cent amplifications observed in field studies conducted by Davis and West (1973). Results of a parameter study indicate topographic effects to be weakly dependent on slope and angle of incidence. However, the effects do change significantly as a function of frequency and position on the scatterer. For receiver positions at depth, scattered body waves were found to be more significant. The radiation lobes of these waves are directed predominantly downward. Qualitative agreement was found between the model results and field results observed on a nearly two-dimensional feature (NASA Mountain near Beatty, Nevada) where underground nuclear explosion ground motions were recorded.