The Bangor Longitudinal Study of Aging (BLSA), conducted in rural Wales from 1979 to 1999, followed a cohort of survivors from more than 500 people over 20 years. Using both quantitative and qualitative data from the study, the factors associated with increases and decreases in loneliness and social isolation were identified. The study was based on a population sample and survivors were followed up every 4 years. From 1983 to 1987, 30 people aged 75 and over in 1979 were studied intensively. The customary measure of loneliness was used, as well as an aggregate measure devised by the research team. Social isolation was similarly measured, using an aggregate measure. Respondents were assessed as demonstrating low, moderate, or high levels of loneliness or isolation. Subsequently, statistical models of loneliness and social isolation were developed. Some respondents were assessed as not experiencing social isolation or loneliness during the study. Others showed changes in levels. In this article, the data are explored, seeking factors associated with changes in social isolation and loneliness. Outcome measures of these two variables of interest are compared with items from the aggregate measures and other identifiable intervening variables. The article discusses which change variables contribute most to levels of isolation and loneliness and result in different combinations of these two outcomes. Implications for policy and practice are discussed.