We investigated the relationship between number of weekly working hours and the prevalence of diabetes in the urban Chinese population. Data regarding anthropometric measurements, fasting blood glucose level and number of hours worked per week were collected from 2228 workers in Shanghai, China (Mage = 44 years; 64% men). Participants were divided into three groups according to the number of hours worked per week (<45, 45–54 and ≥55), and multiple logistic regression analysis was conducted with diabetes as the dependent variable. Subjects with a HbA1c of 6.5% or above or those prescribed anti-diabetic medications were defined as having diabetes. The multivariate adjusted odds ratio for having diabetes was found to be significantly higher for those who worked ≥55 h per week compared with those who worked <45 h per week, but only for men. This finding indicates that working long hours could be a risk factor for diabetes in Chinese male workers. Copyright © 2014 John Wiley & Sons, Ltd.