The 'matching fallacy' suggests that matching schizophrenic patients and normal control subjects on education or IQ may cause systematic mismatching of theoretically expected ability. This study supports a modest version of the matching fallacy effect in nonpsychotic biological relatives of schizophrenic patients. At equivalent levels of education, relatives and control subjects had similar Reading and Spelling scores on the Wide Range Achievement Test-Revised--measures that are largely unimpaired by schizophrenia-related processes. However, relatives showed a deficit in IQ (primarily verbal IQ) compared with what would be predicted from their Reading scores. A similar deficit in Arithmetic scores was found in non-college-educated relatives, but college-educated relatives showed an advantage. We discuss possible implications of the findings with regard to genetic and environmental factors.