The purpose of this study was to examine the association between shift work and diabetes mellitus by separating shift workers according to the intensity of their shift work (seasonal shift work and continuous shift work). Between May and October 2009, we collected data from annual health checkups and questionnaires at a manufacturing company in Shizuoka, Japan. Questionnaires were returned by 1,601 workers (response rate: 96.2%, men/women = 1,314/287). Diabetes mellitus was defined as hemoglobin Alc >= 6.5% and fasting blood sugar >= 126 mg/dl. After exclusions, which included all the women and clerical workers because they did not work in shifts, we analyzed 475 skilled male workers. After adjusting for age, smoking status, frequency of alcohol consumption, and cohabitation status, odds ratios for diabetes mellitus were 0.98 (95% confidence interval [CI]: 0.28-4.81) and 2.10 (95% CI: 0.77-5.71) among seasonal shift workers and continuous shift workers, respectively, compared with non-shift workers. In an age-stratified analysis (< 45 years vs. >= 45 years), the association between continuous shift work and diabetes mellitus was more pronounced among older participants. Compared with non-shift workers, the risk of diabetes mellitus was increased among continuous shift workers, whereas its effect is limited among seasonal shift workers.