Background There is limited knowledge about mortality risk in persons with increased haemoglobin A(1c)(HbA(1c)) levels below the diabetes threshold. Moreover, little is known about how associations between increased HbA(1c)and mortality depend on the length of follow-up. Therefore, we studied associations between HbA(1c)and mortality over long-term follow-up in persons with and without known diabetes. Methods We used data from two German population-based cohort studies: KORA S4 Study (Southern Germany, n = 1458, baseline visits in 1999 to 2001, baseline age 55 to 74 years, mortality follow-up 16.8 years) and Heinz Nixdorf Recall (HNR) Study (Ruhr area, n = 4613, baseline visits in 2000 to 2003, baseline age 45 to 75 years, mortality follow-up 17.8 years). Adjusted log-linear models were fitted to estimate relative risks (RRs) with 95% confidence intervals (CI). Results In both cohorts, participants with HbA(1c)39 to 41 mmol/mol (5.7%-5.9%) and HbA(1c)42 to 46 mmol/mol (6.0% to 6.4%) did not have a larger overall mortality risk than participants with HbA(1c) < 39 mmol/mol (5.7%): the corresponding adjusted RRs were 1.00 (95% CI: 0.83-1.21) and 1.01 (0.80-1.27) in KORA and 0.99 (0.82-1.21) and 0.83 (0.65-1.07) in the HNR Study. For the pooled cohorts, the RR for HbA(1c)39 to 46 mmol/mol (5.7%-6.4%) was 0.96 (0.85-1.07). Associations between newly detected diabetes (HbA(1c) >= 6.5%) and mortality were weak after 4 and 8 years of follow-up, but were stronger after 12 years of follow-up, whereas associations between previously known diabetes (baseline) and mortality decreased. Conclusions HbA(1c)-defined pre-diabetes is not associated with overall mortality. For newly detected and previously known diabetes, mortality risks vary with length of follow-up.