Aims/hypothesis: Screening programmes for type 2 diabetes inevitably find more people at high risk of developing diabetes than people with undiagnosed prevalent diabetes. We describe the incidence of diabetes for risk groups according to advancement in a screening process. Methods: In 2001–2006, a diabetes screening programme based on the Danish diabetes risk score and measures of HbA1c and glucose was carried out in Danish general practices. The present study includes 13,249 individuals with low diabetes risk scores and 22,726 with high diabetes risk scores but no diabetes according to WHO 1999 criteria. Seven incremental levels of diabetes risk were defined and followed for incident diabetes recorded in the Danish National Diabetes Register until December 2012. For each group, cumulative diabetes incidence was calculated. Incidence rates and rate ratios were estimated by Poisson regression analyses. Results: After 10 years of follow-up 1,164 new diabetes cases were registered. Incidence rates were 1.0, 4.2, 14.5, 28.8 and 52.6 per 1,000 person-years in individuals at low risk and in those with normal glucose tolerance, impaired fasting glucose, impaired glucose tolerance and one diabetic glucose value, respectively. For each step in the screening algorithm, the risk of developing diabetes was higher than in the previous step. Conclusions/interpretation: The risk of developing clinical diabetes in people who screen negative for diabetes depends on the level of risk stratification at screening, even at lower risk levels. This risk increases markedly in the presence of impaired glucose regulation. These results can inform policy recommendations concerning prevention strategies following screening. © 2016, Springer-Verlag Berlin Heidelberg.