The purpose of this study was to investigate the issue of age-limited learning effects on visual lexical decision in normal and pathological aging, by using words with different frequency trajectories and cumulative frequencies. We selected words that objectively changed in frequency trajectory from an early word count (Thorndike, 1921, 1932; Thorndike & Lorge, 1944) to a later one (Francis & Kucera, 1982; Kucera & Francis, 1967): we used "Dated words," which changed from high frequency values, in effect during the participants' childhood, to low values, later in development, and "Contemporary words," which showed the reverse pattern of low frequency values in childhood to high ones later in life. Cumulative frequency for these words was assumed to be greater in Contemporary relative to Dated words, as the frequency values from the later word count (1967/1982) remained similar up until testing time according to a current frequency table (CELEX database; Baayen, Piepenbrock, & Gulikers, 1995). As predicted, when word processing involves consistent mappings, no frequency trajectory effect was found in lexical decision. However, a cumulative frequency effect was observed, reflecting an advantage for Contemporary words in both healthy older adults (Experiment 1) and patients with Alzheimer's disease (Experiment 2). This advantage was found to be modified by education level and vocabulary knowledge in healthy older adults (Experiment 3). These participants also performed a subjective AoA rating of the stimuli, which further confirmed that Dated words were acquired before Contemporary words (Experiment 4). Methodological factors are discussed that may account for the finding of age of acquisition effects in lexical decision and other tasks, under some conditions and not others.