This study provides further evidence that the time-of-day (TOD) when testing is conducted affects cognitive performance in old rats and, for the first time in an animal model, in young adult rats as well. Groups of young and old rats were entrained to a 12-h light-dark schedule and administered tests of learning and memory in a non-matching-to-sample (NMTS) task in a water maze. Testing was conducted at the beginning of the rats' activity cycle (AM) or at the end of the cycle (PM). In addition to age differences in performing the task, there were major findings with respect to time of testing: (1) young rats tested in the PM were better than young rats tested in the AM at learning the NMTS rule and in the delayed-NMTS (DNMTS) task; (2) old rats tested in the AM were better than PM-tested old rats on the DNMTS task, with the former attaining performance levels that approximated those of young rats; (3) the TOD effect in old rats extended to a DNMTS reversal (DNMTS-R) condition in which rats, originally tested in the AM, subsequently were administered the test in the PM, and vice versa; (4) the TOD effects in young and old rats in the DNMTS and DNMTS-R tests were strongest at relatively long delays, suggesting that hippocampal function may be particularly vulnerable to such effects; (5) there was evidence in the old rats of a relationship between diurnal drinking patterns and performance at the longest delay in the DNMTS test. These results, which parallel similar findings with human subjects, emphasize a linkage between circadian rhythmicity and cognitive performance throughout adulthood, and indicate the importance of circadian disruption in old age as a contributing factor to age differences in learning and memory performance.