N-Acetyl-aspartate (NAA), a marker of neuronal integrity, has been found to be reduced in frontal regions in schizophrenia. However, the impact of antipsychotic drug type on NAA has not been carefully evaluated. We studied outpatients with schizophrenia/schizoaffective disorders chronically treated with haloperidol or clozapine and normal controls with single-voxel 1H-MRS of the caudate nuclei and the left frontal lobe. Concentrations of NAA, choline containing compounds (Cho) and creatine plus phosphocreatine (Cre) were determined and corrected for the proportion of cerebrospinal fluid (CSF) in each voxel. The haloperidol-treated group had significantly lower CSF-uncorrected and CSF-corrected left frontal NAA than the normal controls, with the clozapine group having intermediate concentrations. The haloperidol-treated group had significantly lower CSF-uncorrected caudate NAA than the normal controls, but the three groups did not differ after correcting for CSF fraction. Performance times in the Grooved Pegboard, a measure of motor dexterity and proxy for parkinsonism, were correlated with CSF-uncorrected and CSF-corrected left frontal NAA. Demographic and illness-related variables were not related to NAA. Exposure to haloperidol-like drugs may in part account for the frontal NAA reductions previously reported in schizophrenia. Adjustment for proportion of voxel CSF should be considered in 1H-MRS studies.