Aims/hypothesis<br>The aim of this study was to assess shift work in relation to incident type 2 diabetes in African-American women.<br><br>Methods<br>In the Black Women’s Health Study (BWHS), an ongoing prospective cohort study, we followed 28,041 participants for incident diabetes during 2005–2013. They answered questions in 2005 about having worked a night shift. We estimated HR and 95% CIs for incident diabetes using Cox proportional hazards models. The basic multivariable model included age, time period, family history of diabetes, education and neighbourhood socioeconomic status. In further models, we controlled for lifestyle factors and BMI.<br><br>Results<br>Over the 8 years of follow-up, there were 1,786 incident diabetes cases. Relative to never having worked the night shift, HRs (95% CI) for diabetes were 1.17 (1.04, 1.31) for 1–2 years of night-shift work, 1.23 (1.06, 1.41) for 3–9 years and 1.42 (1.19, 1.70) for ?10 years (p-trend?<?0.0001). The monotonic positive association between night-shift work and type 2 diabetes remained after multivariable adjustment (p-trend?=?0.02). The association did not vary by obesity status, but was stronger in women aged <50 years.<br><br>Conclusions/interpretation<br>Long duration of shift work was associated with an increased risk of type 2 diabetes. The association was only partially explained by lifestyle factors and BMI. A better understanding of the mechanisms by which shift work may affect the risk of diabetes is needed in view of the high prevalence of shift work among workers in the USA.