An excess of mixed-handedness in schizophrenia has been reported. However, it is not established whether this excess is manifest in non-schizophrenic psychoses, nor whether the underlying etiology is genetic or environmental. We investigated these issues in a group of patients with schizophrenia (n=94), affective psychosis (n=63), other psychosis (n=26); their respective first-degree relatives (total n=183) and a control group (n=85). A narrow definition of mixed-handedness was used corresponding to groups 5 and 6 as defined by the Annett Handedness Questionnaire. We found an excess of mixed-handedness in the schizophrenic group compared with controls (OR=5.2, 1.4-18.6, p<0.006). There was no difference between the other psychotic groups and controls. There was a trend for an excess of mixed-handedness in the first-degree relatives (n=99) of schizophrenic patients (p=0.055), but not in the relatives of affective or other psychotic patients. There was a striking linear trend in the proportion of mixed-handedness between controls, the relatives and the schizophrenic patients (chi2=7.0, p=0.008). There was no association between mixed-handedness and a history of pregnancy or birth complications in the schizophrenic group. There was some evidence for impaired sociability in the mixed-handed schizophrenic patients. Our results indicate that the excess of mixed-handedness in schizophrenia may have a genetic basis.