To examine the effects of cobalamin repletion on cognition in elderly subjects with low serum cobalamin and evidence of cognitive dysfunction. Time series data collected in an open trial of parenteral cobalamin therapy. Outpatient geriatric assessment centers, inpatient geropsychiatry unit, and tertiary care university hospital. Twenty-two subjects with low serum cobalamin (less than 150 pmol/L) and evidence of cognitive dysfunction were entered consecutively over an 8-month period of time. Eighteen subjects completed the study. Subjects received 1000 micrograms of cyanocobalamin intramuscularly daily for 1 week, then weekly for 1 month, then monthly thereafter for a minimum of six months. OUTPATIENT MEASURE: The Mattis Dementia Rating Scale (DRS) was administered both before and at least 6 months after full cobalamin replacement therapy. The hypothesis that cognitive improvement was dependent on the duration of cognitive symptoms was formulated a posteriori. After a minimum of 6 months of cobalamin therapy, 11 of 18 patients showed cognitive improvement. There was a striking correlation between duration of cognitive symptoms and response to therapy. Patients symptomatic for less than 12 months gained an average of twenty points on the DRS (paired t test P = 0.0076), whereas patients symptomatic greater than 12 months lost an average of three points (paired t test P = .34). Two patients symptomatic for only 3 months normalized their DRS scores, gaining 31 and 28 points, respectively. There may be a time-limited window of opportunity for effective intervention in patients with cognitive dysfunction and low serum cobalamin.