The authors reexamined a suspected difference in the effects of clinical information upon detection of nodules vs. more diverse lesions by directly incorporating classification specificity into detection ROC analysis. Categorical prompts, correct for specific abnormalities, led to detection superior to unprompted reading when non-nodule trials (various lesion types) were analyzed. Trials that contained pulmonary nodules, or contained no lesions but had been preceded by "possible tuberculosis" or "rule out metastatic disease" prompts, failed to demonstrate the diagnostic prompt superiority. Perceptual responses may differ when nodules are compared with more complex lesions.