In the present study, F-344 rats throughout 1.5 to 26 months of age were tested in the reference memory version, a moving-platform repeated acquisition version, and in a cued platform version of the Morris water maze. The results suggest that: (1) performance in the water maze declines continuously, beginning at the earliest age, and very closely fits a linear function; (2) there are robust, reliable differences between individuals in terms of their performance in the Morris water maze, but chronological age accounts for only a fraction of the variance between individuals; (3) there is no evidence of a bimodal distribution among aged rats--there is no distinct subgroup of individuals that performs so poorly that they are qualitatively different from the majority of the population, and distinctions between "impaired" and "unimpaired" subjects must be based on arbitrary criteria that may not be consistent from one study to the next; (4) age-related deficits in the Morris water maze may not be restricted to learning and memory, but may also include deficits in attention, the ability to process spatial information, and/or the ability to develop efficient spatial search strategies; and (5) swim distance is the most appropriate measure of cognitive function in the Morris water maze, but the relationship between this measure and other measures of noncognitive function make it clear that swim distance may not be a pure measure of cognitive function. Although the Morris water maze remains a valuable preclinical test with better validity and specificity than many other behavioral tests, measures of performance in the Morris water maze should not be considered synonymous with cognitive function.