Spatial working memory and planning abilities were assessed in 36 hospitalized patients with chronic schizophrenia, using the computerized Cambridge Neuropsychological Test Automated Battery (CANTAB), and compared with those of normal subjects and patients with neurological disorders (frontal lobe lesions; temporal lobe and amygdalohippocampal lesions; Parkinson's disease), matched for age, sex and National Adult Reading Test IQ. The patients in the group with temporal lobe lesions were unimpaired in their performance on these tasks. Patients with schizophrenia were impaired on visuo-spatial memory span compared with all the other groups, while severity of Parkinson's disease was correlated with the degree of impairment on this task. The patients with schizophrenia and those with frontal lobe lesions were impaired on a 'spatial working memory' task, with increased 'between-search errors'. Patients with Parkinson's disease performed this task poorly compared with the younger control subjects. Patients with schizophrenia were unable to develop a systematic strategy to complete this task, relying instead on a limited visuo-spatial memory span. Higher level planning ability was investigated using the CANTAB 'Tower of London'. All groups were equally able to complete the task. However, the groups of patients with schizophrenia and frontal lobe lesions made fewer perfect solutions and required more moves for completion. Movement times were significantly slower in the schizophrenia group, suggesting impairment in the sensorimotor requirements of the task. The patients with schizophrenia were not impaired in their 'initial thinking' (planning) latencies, but had significantly prolonged 'subsequent thinking' (execution) latencies. This pattern resembled that of the group with frontal lobe lesions and contrasted with the prolonged 'initial thinking' time seen in Parkinson's disease. The results of this study are indicative of an overall deficit of executive functioning in schizophrenia, even greater than that seen in patients with frontal lobe lesions. However, the pattern of results in schizophrenia resembled that seen in patients with lesions of the frontal lobe or with basal ganglia dysfunction, providing support for the notion of a disturbance of frontostriatal circuits in schizophrenia. Our findings also indicate that there is a loss of the normal relationships between different domains of executive function in schizophrenia, with implications for impaired functional connectivity between different regions of the neocortex.