In order to understand the angular dependence of reconstructed images, a "hot rod" phantom was simulated with six rod sizes ranging from 12.7 mm to 4.8 mm, arranged in 60 degree sectors. This simulation was used to evaluate changes caused by increasing the number of views from 60 to 180 and to compare results to published angular sampling requirements. Also, a comparison was made of step-and-shoot and continuous data acquisition modes. Experimental phantom data were taken for three acquisition modes and for two levels of image statistics and compared with the simulation. This study shows the effects of angular aliasing and that image quality generally improves with increasing views even if total image counts are reduced. Despite smoothing effects, a continuous (180 view) scan is, in most cases, the preferred acquisition mode because it offers the highest angular sampling, as well as maximum counting efficiency.