The use of the aged mouse as an integrated model of age-related cognitive decline is reviewed, with special emphasis on experiments covering the life span of NMRI mice, using different age-groups ranging from 3 through to 22 months. Age-related changes in the sensorimotor profile, spontaneous behaviour and performance in learning and memory tasks are considered. The data provide evidence for cognitive impairment and decreases in spontaneous activity and exploration from middle age onwards. Chronologically, this age depends on the longevity of the strain selected; in NMRI mice, middle age corresponds to 11-12 months. Complex learning tasks, such as the Morris water maze for spatial learning, appear to be the most sensitive to age-related changes, as are tests requiring prolonged retention of acquired information, for example, using passive avoidance. Cued and simple discrimination learning are only impaired in the oldest animals. Age-related changes in non-cognitive variables, including sensorimotor capacity, pain sensitivity, emotionality, or locomotor activity, do not account for the learning impairments, although deficits in visual acuity cannot be excluded in the very old animals. Detailed analysis of the individual data for middle aged and old mice, using discriminant and correlation studies highlight a marked heterogeneity between animals of any given chronological age. Furthermore, individual aged mice do not exhibit similar degrees of impairment across all the behavioural variables, showing that aging is not a uniform process. The possible relationship between age-related behavioural decline and neurochemical changes is an area as yet unexplored apart from a few isolated investigations, including a study on ChAT and AChE in NMRI mice. The studies in the NMRI mice illustrate the value of investigating the full age-range to detect an age group which shows cognitive decline dissociable from physical or emotional changes and which is representative of the population as a whole.