Depressed geriatric patients show substantial intersubject variability in cognitive performance, which complicates attempts to evaluate the cognitive effects of depression and of antidepressant therapy. This variability may reflect the multiple medications older patients take, many of which have anticholinergic effects. This study examined whether serum anticholinergicity (SA) explained some of the variability in depressed geriatric patients' memory performance. Before starting antidepressant treatment, 36 elderly depressed subjects were given a verbal learning test. At the same time, a blood sample was taken and analyzed by radioreceptor binding assay to determine their SA level. Nineteen of the subjects had detectable levels (mean = 0.28 pmole atropine equivalent). Subjects with an SA of zero showed significantly better delayed recall than did those with a positive SA level. Thus, even very low SA may produce subtle decrements in memory performance, an area of cognition known to be highly sensitive to anticholinergic effects.