A total of 18 outpatients (17 male, 1 female) ranging in age from 36–66 years old were on a constant dosage of haloperidol in equally divided doses at 9:00 a.m. and 9:00 p.m. for at least 1 month. DSM-III-R diagnoses included schizophrenia (N = 9), schizoaffective disorder (N = 3), bipolar disorder (N = 4), organic mental disorder (N = 1), and delusional disorder (N = 1). Blood samples for steady-state concentrations of plasma and red blood cell haloperidol (H) and reduced haloperidol (RH) were drawn at 9:00 a.m. (12 hr trough). The haloperidol dosage was held at 9:00 a.m. until samples of whole saliva and parotid saliva could be collected for flow rates and concentrations of H and RH. Haloperidol dosages ranged from 1 mg/day to 60 mg/day (mean 11 ± 15). Correlation coefficients were calculated for saliva concentrations versus blood levels and for saliva secretion rates versus blood levels. The correlations between whole saliva measures and blood concentrations were all higher than the correlations between parotid saliva measures and blood concentrations. In one case the higher correlation reached statistical significance. There was only one case in which substitution of saliva secretion rate improved the correlation between measures with saliva concentration. Our findings suggest that saliva measures of H and RH are useful alternatives to plasma concentrations in monitoring maintenance haloperidol treatment.