Although automated continuous performance tests (CPT) are gaining popularity as aids to the diagnosis of attention deficit disorder (ADD), little is known of their validity in this context. Our preliminary experience with a commercially available visual CPT indicated that as many as a third of children meeting the DMS-III criteria for ADD may score well enough on this measure to escape detection. We therefore analyzed the results of neuropsychological testing as well as CPT performance in 14 ADD children and six non-ADD children in an effort to determine whether CPT performance might reflect higher level cognitive variables other than attention and/or impulsivity. We found that those ADD children classified as "abnormal" on the basis of the CPT scored significantly below those classified as "normal" on measures of abstract reasoning and logical problem solving, simple verbal reasoning, nonverbal problem solving, and simple arithmetic skills. The non-ADD group contained a high proportion (83%) of subjects with CPT performance outside of the normal range. These data suggest that CPT may yield both false negative and false positive results when used as screening tools for ADD, and we recommend therefore that caution be used in their interpretation.