The aim of the study was to search for bivariate associations between on the one hand samples of background, job-related and well-being characteristics, and on the other hand longitudinally recorded sick-leave in individuals with repeated short-term sick-leave. Out of altogether 123 subjects included, 113 self-rated their well-being at the onset of, and 88 also after, a supportive programme. The well-being rating covered experience of treatment by other people, reservation, loneliness, inferiority, tension, vulnerability, guilt, security and indolence. Socio-demographic and job-related characteristics and attitudes were registered and declared at the initial contact. Correlations between these separate sets of independent variables, and sick-leave occasions and days in each of the two years following the contact were computed. There was an almost complete lack of covariation between background variables and sick-leave. Neither did job-related characteristics correlate. Only an expressed wish for a job change was vaguely associated with more absence days, but not with more spells. In striking contrast, a convincing consistent longitudinal pattern between self-rated well-being and sick-leave behaviour emerged. The poorer the well-being, the greater the subsequent sick-leave. Yet, a major part of the sick-leave variance remained unexplained. Accordingly, other etiological factors may still interact with well-being in the comprehensive setting behind the sick-leave behaviour.(ABSTRACT TRUNCATED AT 250 WORDS)