Two experiments addressed the utility of a T-maze, delayed reversal paradigm for assessment of recent memory impairment in aging C57BL/6NNia mice. This paradigm involved acquisition of a learning set for identification of the correct arm choice in a daily multiple-trial discriminated avoidance session. During each session, the correct arm was always opposite that entered on an initial 'information trial' and maximal performance thus required a reversal (or 'lose-shift') strategy. Once the learning set had been acquired, retention performance was examined following delays of varying length which were introduced following the information trial during each session. In the first experiment, acquisition and retention components of delayed reversal were considered in a cross-sectional study involving separate groups of mice aged 7, 10, 15, or 27 months. Analysis of acquisition components suggested that relative to young mice, the old mice were slower to acquire both the first reversal and the learning set. Analysis of the retention phase suggested that memory decay gradients for goal arm discrimination became more pronounced with increasing age, whereas decay gradients for the avoidance response were similar among the age groups. Correlational analysis of data for the old mice suggested independence of age-related deficits in the acquisition and recent memory components of the delayed reversal paradigm. In the second experiment, survivors from the previous 7- and 10-month-old groups were retested when 27 months of age. The cross-sectional and longitudinal data led to the same conclusions with regard to the effect of age on recent memory function. Overall, these results suggest that the delayed reversal paradigm will be a valuable tool in the analysis and evaluation of interventions potentially affecting age-related cognitive impairment.