The main objective of this study was to investigate reading habits in older adults in relation to level and 15-year changes in verbal fluency and episodic recall. We examined a sample of 1157 participants (55 years at baseline) up to 15 years after the baseline assessment using latent growth curve modeling of cognitive measures with baseline reading frequency (books, weekly magazines) as a predictor of cognitive level (intercept) and rate of change (slope). Subgroup analyses were performed to investigate the role of an early adult g factor in the association between reading habits and cognitive ability in midlife. Frequent reading of books, but not of magazines, was associated with higher levels of verbal fluency and recall but unrelated to rate of longitudinal decline. Subgroup analyses indicated that the g factor in early adulthood predicted reading and cognitive level in midlife and this factor removed the current association between reading habits and level of cognitive ability (both cognitive factors). The results indicate an enduring relationship between book reading and level of cognitive ability across the adult life span and provide little support of the hypothesis that frequent reading protects against latelife cognitive decline. The extent to which book reading promotes cognitive functioning in childhood/youth remains to be demonstrated. Intervention studies may be useful in this regard.