Over four experiments based on the delayed matching-to-sample task, fornix-transected and normal control monkeys were presented with a sequence of five sample stimuli and then received intermixed within-session recency (WSR) and between-session recency (BSR) tests in experiment 1, only BSR tests in experiment 2, only absolute novelty (AN) tests in experiment 3, or only WSR tests in experiment 4. In WSR tests, monkeys chose which of two samples had occurred more recently in the immediately preceding sequence. In BSR and AN tests, monkeys were required to choose one sample from the immediately preceding sequence in preference to a foil unseen in the present session (BSR) or an AN foil that had never been presented before. When tests of WSR and BSR were intermixed (experiment 1), fornix monkeys performed below the level of the control monkeys in both types of test, although this difference was not statistically significant. In experiment 2, fornix monkeys were significantly impaired on tests of BSR alone, in which memory for a stimulus presented in an immediately preceding sequence could compete with memory for a foil presented in an earlier training session. In tests of AN (experiment 3), fornix monkeys performed at the same level as control animals in distinguishing a previously experienced stimulus from a previously unseen foil. In experiment 4, fornix transection significantly impaired tests of WSR alone. Taken together, these results suggest that one specialized role of the fornix is to process temporal information.