Spatial memory is of interest in schizophrenia because of widespread impairments in adaptive functioning, including independent living skills. Short-term spatial memory is impaired in this disease, whereas spatial reference memory, a longer-term spatial memory, has not been evaluated. Animal studies have demonstrated that anticholinergics impair short-term spatial memory but not spatial reference memory. The effects of haloperidol and risperidone on these two types of spatial memory were evaluated in a double-blind randomized comparison in inpatients with schizophrenia. It was predicted that risperidone would have a greater beneficial effect on spatial working memory than haloperidol. Computerized measures of spatial working memory and spatial reference memory were developed based on animal assessment of these functions. Subjects with schizophrenia were assessed during a medication-free period and again following 4 weeks of fixed-dose treatment. Risperidone, compared to haloperidol, improved spatial working memory performance, an effect that became nonsignificant when benztropine co-treatment was controlled. There were no treatment effects on spatial reference memory performance. Consistent with animal studies, benztropine impaired spatial working memory but not spatial reference memory. The relative benefits of risperidone on spatial working memory performance were largely explained by differential benztropine treatment for the haloperidol-treated subjects.