Efforts to find a correlation between serum levels of chlorpromazine (CPZ) and clinical effect have been rather unsuccessful, which could be due to fluctuations of CPZ and CPZ metabolite levels during treatment, the complicated metabolism of CPZ, or to varying degrees of protein binding. Using a mass fragmentographic analysis technique the variations of CPZ and two active metabolites nor1-CPZ and 7-OH-CPZ were studied in ten schizophrenic patients during the day at steady state and after withdrawal. There was a significant correlation between the area under curve (AUC) for CPZ in serum during 24-h treatment and serum concentration at different fixed times of the day. The half-life (T 1/2) for CPZ was found to be 8-33 h. 7-OH-CPZ and nor1-CPZ disappeared at about the same rate as the parent compound. The concentration of both metabolites was less than 10 ng/ml after 36 h. CPZ was administered to 43 schizophrenic patients in one of three fixed doses (200, 400, or 600 mg) according to a double-blind design. Plasma and cerebrospinal fluid (CSF) samples were analysed before and after both 2 and 4 weeks' treatment. The levels of the metabolites were considerably lower as compared to CPZ levels. The same levels were found after 2 and 4 weeks' treatment. There were no sex differences. The levels of CPZ and metabolites presented a weak positive correlation to daily dose but not to dose calculated by mg/kg body weight. Older patients tended to have higher CPZ and metabolite levels. The dose effects were very similar in plasma and CSF and there was a highly significant correlation between CPZ levels in serum and CSF. CSF/plasma ratio for CPZ seems to be an individual factor possibly related to variations in protein-binding in plasma and CSF.