It is accepted that spiral CT scanning may offer significant advantages in a number of clinical applications. There is still some concern with respect to image quality, however, since slice sensitivity profiles are slightly broadened due to the table motion. We carried out theoretical analysis, phantom measurements, and computer simulations to evaluate and to compare contrast and spatial resolution for conventional and for spiral scanning. Special emphasis was put on the task of detecting spherical lesions. For standard test objects that measure only resolution in the scan plane, no significant difference between conventional and spiral scanning was observed. We therefore designed a phantom setup that allowed us to place spheres of arbitrary diameter and contrast in arbitrary positions to test three-dimensional (3D) resolution. For conventional CT, both lesion contrast and the degree of spatial separation of lesions observed depend on the relation of the start position of the scan series to the random location of a sphere or lesion. Spiral CT offers space-invariant resolution due to its continuous scanning. Small lesion contrast may be improved by up to a factor of 1.8 when compared with conventional CT since slices can be centered retrospectively. Measurements and simulations were in excellent agreement. We conclude that spiral CT can offer improved 3D contrast and spatial resolution. To exploit these advantages, images should be reconstructed in spiral CT at increments of less than half the distance traveled during one 360° tube rotation. With four to five images per such interval, usually equal to the slice width, results very close to the theoretical optimum are achieved. Many of the presented considerations and results apply to other slice imaging modalities like MRI in an analogous fashion.