A new and syntactically more complex version of the Token Test was developed to investigate the comprehension of complex syntax in schizophrenia. Results suggested that schizophrenic patients as a group demonstrated an impairment of comprehension proportionate to a reduction in the syntactic complexity of their spoken language, which had been demonstrated in two separate studies. While manic patients also recorded poorer results on the new Token Test than control subjects, there were no significant correlations between their comprehension and production measures as there were for the schizophrenic patients. The speculation is made that manic patients may perform poorly because of state factors such as distractibility, while schizophrenic patients may do so because of an underlying language impairment.