Ten patients with biopsy-proven Alzheimer's disease (AD) received low-dose (0.35 mg/d) intraventricular bethanechol, a muscarinic agonist, and saline placebo in a 24-week double-blind crossover design. Eight of these ten patients later participated in an open escalating-dose (to 1.75 mg/d) trial of bethanechol. Patients' drug responses were assessed by neuropsychological examination and informant measures of activities of daily living, mood disturbance, and abnormal behavior. Bethanechol appears to have a narrow therapeutic window for positive effects; low doses did not reliably alter patient functioning, moderately increased doses appeared to have a palliative effect on patient mood and behavior, and the highest dose was detrimental to patient functioning. Bethanechol does not appear to ameliorate the dementia of AD, but may exert a mildly positive effect on patient behavior and mood.