Purpose Work ability is commonly measured with self-assessments, in the form of indices or single items. The validity of these assessments lies in their predictive ability. Prospective studies have reported associations between work ability and sickness absence and disability pension, but few examined why these associations exist. Several correlates of work ability have been reported, but their mechanistic role is largely unknown. This study aims to investigate to what extent individual's own prognosis of work ability predicts labor market participation and whether this was due to individual characteristics and/or working conditions. Methods Self-assessed prognosis of work ability, 2 years from "now," in the Stockholm Public Health Questionnaire (2002-2003) was linked to national registers on sickness absence, disability pension and unemployment up to year 2010. Effects were studied with Cox regression models. Results Of a total of 12,064 individuals 1466 reported poor work ability. There were 299 cases of disability pension, 1466 long-term sickness absence cases and 765 long-term unemployed during follow-up. Poor work ability increased the risk of long-term sickness absence (HR 2.25, CI 95 % 1.97-2.56), disability pension (HR 5.19, CI 95 % 4.07-6.62), and long-term unemployment (HR 2.18, CI 95 % 1.83-2.60). These associations were partially explained by baseline health conditions, physical and (less strongly) psychosocial aspects of working conditions. Conclusions Self-assessed poor ability predicted future long-term sickness absence, disability pension and long-term unemployment. Self-assessed poor work ability seems to be an indicator of future labor market exclusion of different kinds, and can be used in public health monitoring.