Objectives: The present study is an investigation of the association between job stress, determined on the basis of a demand-control model or worksite social support at the baseline, and absence due to illness among employed Japanese males and females. Methods: We analyzed 448 male and 81 female subjects who had taken no sick leave in the year preceding the baseline (1997) and observed them all until 1999. A self-administered questionnaire was the source information collected. It consisted of questions on socio-demographic variables, occupations, health-related behavior, a Japanese version of the Job Content Questionnaire, and the number of absences in the year preceding both the baseline and follow-up. Logistic regression analyses were used to determine how the characteristics of a job at the baseline affected sickness absence of 5 days or longer per year; controls were established for the gender, age, level of education completed, occupation, number of cigarettes smoked daily, and the amount of alcohol consumed weekly. Results: Compared to the lowest tertile of the ratio of demand to control (job strain), the highest tertile was significantly associated with an increased risk of sickness absence of 5 days or longer per year (odds ratio 3.02; 95%CI 1.00-9.16) at follow-up. The dose-response relationship was supported (p for trend <0.05). However, individual variables of job demand, job control, and worksite social support were not significantly associated with the risk of absence from illness. Conclusions: The study provided prospective evidence that job strain leads to an increased risk of sick leave among Japanese employees. © Springer-Verlag 2005.