Nineteen patients with probable Alzheimer's disease who manifested psychosis or behavioral disturbance received 0.5 to 5 mg per day of oral haloperidol, with blood levels drawn on a stable daily dose. Oral dose correlated strongly with haloperidol blood levels assessed by radioimmunoassay (r = 0.82, p = .0001). The stronger correlation observed between oral dose and blood level compared with most studies in schizophrenia may be related to the fact that virtually all the Alzheimer's patients were neuroleptic naive, without the alterations in absorption and metabolism known to occur in schizophrenics receiving long-term neuroleptic treatment. Compared with oral dose, blood levels showed stronger relations to changes in symptoms and to changes in severity of extrapyramidal side effects. These preliminary data suggest that the utility of monitoring haloperidol blood levels in patients with Alzheimer's disease merits further investigation.