In animals, evidence has been accumulated that vasopressin (VP) improves learning and memory. In humans, this effect was not consistently demonstrated, and attempts to restore age-related memory deficits by VP also remained inconsistent. Assuming that in old subjects a beneficial effect on memory occurs only after prolonged treatment with VP, we conducted a study in 26 healthy elderly persons receiving 40 IU of VP for three months through the intranasal route. The trial was randomized, placebo-controlled and held double-blind. Memory was assessed by the Auditory Verbal Learning Test (AVLT) requiring the subject to learn repeatedly presented lists of 15 words. Results demonstrated no general effect of long-term treatment with VP on memory in aged humans. However, recall of an interfering word list was improved, indicating a diminished proactive interference by the peptide. Additionally, VP influenced recall depending on the serial position of an item: it improved the primacy effect (i.e. recall of the first words of a list) and impaired the recency effect. This result may indicate an improved semantic encoding (i.e. a primary effect on processes of attention) after long-term administration of VP.