Research has long demonstrated that loneliness is a key risk factor for poor health. However, less isknown about the development and predictors of loneliness across later adulthood. We examined thesequestions using two-wave data obtained 5 years apart in the population-based Norwegian NorLAG study(N 5,555; age 40–80 years; 51% women). We considered direct measures of loneliness (askingdirectly about feeling lonely) and indirect measures (avoiding the term loneliness) and linked them toself-report data on personality and contact with friends, and to register data on socioeconomic (education,income, unemployment), physical health (sick leave, lifetime history of disability), and social factors(children, marriage/cohabitation, lifetime history of divorce and widowhood). Results indicated thatlevels of loneliness increased steadily for women, whereas men’s levels followed aU-shaped curve, withhighest loneliness at ages 40 and 80. At age 40, loneliness declined between the two data waves, but withincreasing age the decrease abated and turned into increases when loneliness was measured indirectly.Disability, no spouse/cohabiting partner, widowhood, and little contact with friends were each associatedwith more loneliness. Similarly, people high in emotional stability and extraversion reported lessloneliness and experienced steeper loneliness declines on one or both loneliness measures. We take ourresults to illustrate the utility of combining self-report and register data and conclude that the develop-ment of loneliness across the second half of life is associated with both individual difference character-istics and aspects of social embedding. We discuss possible mechanisms underlying our findings andconsider practical implications.