Studies on the withdrawal of anticholinergics from patients on antipsychotics have produced conflicting results. This 12-week study employed a double-blind crossover design on 39 adult in-patients selected from a total hospital population of 620. The Colombia Scale was used to determine extrapyramidal side effects (EPS). All patients were stabilised prior to the study on benztropine mesylate 2 mg b.i.d., and gradual withdrawal was employed. Benztropine withdrawal produced a significant increase in overall EPS scores. Ten patients (26%) required reinstatement of benztropine while on placebo. Sialorrhoea, rigidity and postural instability were the most prominent changes. Neither age, sex, nor diagnosis were significantly predictive of EPS. Depot medications and doses greater than 1000 mg/day chlorpromazine-equivalent were related to significant EPS increase. The intrinsic anticholinergic properties of the antipsychotics themselves and concomitant medications, such as antidepressants, appeared protective against development of EPS. Most patients on a combination of antipsychotics and anticholinergics can safely be withdrawn from the latter.