Haloperidol has been used extensively for the treatment of psychotic disorders, and it has been suggested that the monitoring of plasma haloperidol concentration is clinically useful. Different assay methodologies have been used in research and clinical practice to examine the relationship between response and plasma concentration of the drug. Chemical assays such as high pressure liquid chromatography (HPLC) and gas-liquid chromatography (GLC) have good precision and sensitivity; radioimmunoassay (RIA) is generally more sensitive, but less precise and specific. Radioreceptor assay quantifies dopaminereceptor blocking activity but does not provide results comparable with those of HPLC, GLC and RIA. Large doses of haloperidol can safely be given intravenously and intramuscularly for rapid neuroleptisation; the bioavailability of this agent administered orally ranges from 60 to 65%. However, there is large interindividual, but not intraindividual, variability in plasma haloperidol concentrations and most pharmacokinetic parameters. This interindividual variability could be partially explained by the reversible oxidation/reduction metabolic pathway of haloperidol: it is metabolised via reduction to reduced haloperidol, which is biologically inactive. Different extents of enterohepatic recycling, and ethnic differences in metabolism, could also account for the observed variability in haloperidol disposition. Although not conclusive from different clinical studies, it appears that a plasma haloperidol concentration range of 4 micrograms/L to an upper limit of 20 to 25 micrograms/L produces therapeutic response. The role of reduced haloperidol in determining clinical response is not clear, although in some studies a high reduced haloperidol/haloperidol concentration ratio has been suggested to be associated with therapeutic failure. Measurements of red blood cell or cerebrospinal fluid haloperidol concentration have also been proposed as determinants of therapeutic response, but results from different studies are inconsistent, and do not seem to provide a significant advantage over plasma concentration monitoring. Physiological parameters such as prolactin and homovanillic acid levels have been evaluated, with the latter showing some promise that warrants further investigation. Haloperidol decanoate can be characterised by a flip-flop pharmacokinetic model because its absorption rate constant is slower than the elimination rate constant. Its plasma concentration peaks on day 7 after intramuscular injection. The elimination half-life is about 3 weeks, and the time to steady-state is about 3 months.