SUMMARYThirty-six female inpatients with a clinical diagnosis of senile dementia of the Alzheimer type were entered into a double-blind withdrawal of their established thioridazine therapy over a four-week period. Prolactin (PRL) levels were assayed at weeks 0, 2 and 4 in both groups. A significant reduction in PRL level over this period was seen in the placebo-substituted group as compared with the active-continued group. Over the same interval there was no significant change or difference between the two groups in terms of cognitive function, behaviour or physical condition. There was no correlation between PRL levels and degree of dementia or dose of thioridazine at the start of the study. This study suggests that the response to thioridazine withdrawal in senile dementia of the Alzheimer type is a reversal of pharmacological hyperprolactinaemia. The timescale and extent of this can be likencd to that seen in schizophrenics and in normal volunteers. No definite evidence emerged that PRL levels might be useful as markers of disease severity or response to neuroleptic therapy.KEY woms-Thioridazine, prolactin, dementia.In a previous publication (Findlay et al., 1989) we have reported a study into the behaviour of thioridazine treatment in a population of elderly female inpatients with a diagnosis of senile dementia. Over the period of the study, there were no withdrawal phenomena, and although there was a trend which showed a decline in side-effects for the group in which thioridazine was discontinued, the difference from the group in which treatment was continued was not statistically significant. This suggested the controlled withdrawal of thioridazine was safe, and should be considered in situations where the value of continued medication was in doubt.Thioridazine, along with other phenothiazines, blocks postsynaptic dopamine (DA) receptors in the tuberoinfundibular system of the hypothalamus and anterior pituitary, thus preventing DA mediated inhibition of PRL release (Meltzer and Fang, 1976;Sachar et al., 1975;Beaumont et al., 1974). Neuroleptics are widely used in the treatment of the major psychoses, and also to control behavioural problems in patients suffering from dementia (Barnes et al., De la Fuente and Rosenbaum, 1981).There have been many studies carried out examining neuroendocrine function in dementia, and