Although a number of analyses have addressed whether initial cognitive ability level is associated with age-related cognitive decline, results have been inconsistent. Latent growth curve modeling was applied to two aging cohorts, extending previous analyses with a further wave of data collection, or as a more appropriate analytical methodology than used previously. In the Lothian Birth Cohort 1921, cognitive ability at age 11 was not associated with cognitive change from age 79 to 87, either in general cognitive ability, or in tests of reasoning, memory and executive function. However, data from the MRC National Survey of Health and Development suggested that higher cognitive ability at age 15 predicted less decline between ages 43 and 53 years in a latent cognitive factor from tests of verbal memory and search speed, and in search speed when considered separately. The results are discussed in terms of the differences between the cohorts and the interpretability of the analytical approach. Suggestions are made about when initial ability might be cognitively protective, and study requirements to bring about a clearer resolution. (Contains 2 figures and 7 tables.)