Objectives and methods • The aim of this research is to investigate the effects of adult learning upon a range of measures of health and social capital and cohesion. • The study has a broad scope. We estimate effects of adult learning on a wide range of outcomes. We break down participation in learning in a number of ways, and we assess the different impacts of participation on different groups. The primary objective is to assess the nature and extent of wider effects of adult learning. More detailed findings relating to specific outcomes, types of courses taken and groups of learners are extremely interesting and indicate directions for further research. • The National Child Development Study is the database used for analyses. We use data relating to almost 10,000 adults born in Britain in 1958, focusing upon changes in their lives between the ages of 33 (in 1991) and 42 (in 2000). • The effects of participation in learning between the ages of 33 and 42 are estimated. Participation in learning is broken down according to the type of course taken – courses resulting in academic and vocational qualifications, work-related training and leisure courses. • Previous, closely related research on the effects of adult learning focused on wage effects only and neglected non-accredited learning of the kind considered here. The current project is, therefore, a timely reminder that many learners take courses in adult learning that are not accredited and for reasons other than wage returns. A proper assessment of the value of adult learning must take these wider benefits into account. • The outcomes considered are changes between age 33 and 42 in social and political attitudes, civic participation, health behaviours (smoking, alcohol consumption and exercise), self-reported life satisfaction and onset and recovery from depression. • Effects of participation upon these outcomes are estimated for the whole sample. In addition, the different effects of participation are separately estimated for men and women, and for those without Level 2 qualifications at age 33. • We restrict selection bias by exploiting the fact that our data are longitudinal. We attempt to consider effects on individuals as they change over time, making before and after comparisons. We also control for a wide range of life-course background, achievement and developmental factors.