The authors respond to issues raised about data from the Victoria Longitudinal Study and further explain concerns regarding evidence for the engagement hypothesis. Discussion focuses on the use of social stratification variables such as occupational prestige and educational attainment as measures of an engaged lifestyle. It is argued that (a) tests of the hypothesis should focus on the relationship of behaviors and activities thought to be proximal beneficial influences on adult cognitive development; (b) persuasive evidence for engagement effects from existing data require demonstration of effects of intellectual activities that are statistically independent of associations of social status with intellectual and cognitive development; and (c) currently available longitudinal data do not provide definitive evidence regarding the benefits of an engaged lifestyle on cognitive change.The belief that staying mentally active during adulthood can promote or enhance intellectual and cognitive function in late life is widely held by laypersons and scientists alike. According to Arbuckle, Maag, Pushkar, and Chaikelson (1998), an engaged adult lifestyle is associated with a more complex social environment, in which "intellectual abilities are more likely to be challenged, practiced, and reinforced, resulting in the development and maintenance of higher levels of intellectual ability in later life" (p. 663). Recently, we (Hultsch, Hertzog, Small, & Dixon, 1999) used a structural equation model (SEM) with latent change factors to determine whether activity patterns were related to level and change of performance on a wide variety of cognitive tasks in data from the Victoria Longitudinal Study (VLS). Although engagement in activities thought to be intellectually demanding (e.g., reading novels, playing bridge) did relate to 6-year longitudinal changes in cognition in our sample of older adults (ages 55 to 84 at initial test), we also observed that there were strong concurrent changes in both cognition and activity levels. This finding led us to estimate an alternative model, one in which changes in cognition caused reduced intellectual activity levels in older adults.That both models provided reasonable representations of the data led to a major shift in our thinking. Like others in the field, we entered the analysis with the prior expectation that the stream of causal influence ought to extend from an engaged lifestyle and cognitively demanding activities to changes in cognition. Our analysis led us to conclude, however, that the field had not adequately addressed the