Thirty-six female inpatients with a clinical diagnosis of senile dementia, Alzheimer type, were entered into a doubleblind withdrawal of their established thioridazine. Over a four-week period there were no significant differences between the two study groups in terms of cognitive function, behaviour or physical state. There was no evidence of a withdrawal phenomenon and the use of replacement medication was at a similar low level in the two groups. Although significant differences were not established in terms of side-effect, there was a trend for greater reduction of these for the group in whom treatment was stopped. These data indicated that careful, controlled withdrawal of established thioridazine treatment is a safe procedure which may be useful in circumstances where the contribution of medication is unclear.