A test predictive of ultimate radiologic expertise could be of great value in the selection of individuals entering the field. Some individuals have an aptitude superior to others to perceive three-dimensional spatial relations from two-dimensional data. This may enhance their ability to draw radiologic conclusions from clinical images and may favorably affect their performance as radiologists. To test cognitive perceptual ability, a three-dimensional Visual Form Reconstruction Test (form test) was developed and administered to residents and faculty members. All subjects also completed the Thurstone Surface Development Test, a standardized test of spatial visualization ability. The form test results correlated well with resident performance, as measured by overall faculty ratings (predictive validity). Although form test and Thurstone test performance were highly correlated with each other (concurrent validity), the form test was better correlated with resident performance. Intercorrelations among the three subsections of the form test demonstrated high split-part reliability. Performance on the form test was unaffected by level of training. This suggests that an underlying aptitude was measured. These preliminary results indicate that testing of spatial visualization aptitude is predictive of resident performance in radiology. A test such as this could be useful in selection and self-selection of resident candidates.