AimTo evaluate the relationship between thyroid-stimulating hormone (TSH) levels within the normal range and the risk of type 2 diabetes mellitus (T2DM) in a cohort of patients at high cardiovascular risk, and to perform a systematic review and meta-analysis of previous studies.MethodsWe included 5542 patients without T2DM from the prospective Secondary Manifestations of ARTerial disease study with TSH levels between 0.35 and 5.0 mIU/L without anti-thyroid medication or thyroid-hormone replacement therapy. Cox regression was used to investigate the relationship between baseline plasma TSH levels and incident T2DM. MEDLINE, EMBASE, and Cochrane were searched for prospective cohorts assessing TSH and incident T2DM. Hazard ratios (HR) from included prospective cohort studies were pooled using a random-effects model.ResultsIn patients at high cardiovascular risk, higher plasma TSH levels in the normal range were not associated [HR 1.07 per mIU/L increase in TSH (95% confidence interval (95% CI) 0.95–1.22)] with an increased risk of T2DM, adjusted for age, sex, smoking, total and HDL cholesterol, and triglycerides. In the meta-analysis involving three prospective cohort studies, including the present study, including 29,791 participants with 1930 incident events, there was no relation between plasma TSH levels in the normal range and incident T2DM [pooled HR 1.06 (95% CI 0.99–1.14)].ConclusionThere is no apparent relation between plasma TSH levels in the normal range and incident T2DM in patients at high cardiovascular risk.Electronic supplementary materialThe online version of this article (10.1007/s00592-018-1231-y) contains supplementary material, which is available to authorized users.