The prognostic importance of baseline and serial glycated hemoglobin (HbA1c) changes for cardiovascular outcomes is still debated. We aimed to evaluate it in 620 high-risk individuals with type 2 diabetes (mean age 60.4 years, 37 % males, 55 % Caucasians). Patients had HbA1c levels measured at study entry and serially during follow-up. Primary end points were total cardiovascular events (CVEs), major CVEs (non-fatal myocardial infarctions and strokes plus cardiovascular deaths) and all-cause mortality. Cardiovascular and non-cardiovascular mortalities were secondary end points. HbA1c was evaluated either as a continuous variable and categorized at clinically relevant cutoffs. Multivariate Cox regressions assessed the associations with end points. After a median follow-up of 6.6 years, 125 total CVEs occurred (90 major CVEs), and 111 patients died (64 from cardiovascular diseases). After statistical adjustments for other cardiovascular risk factors, baseline and mean first-year HbA1c predicted all end points, except non-cardiovascular deaths; and hazard ratios tended to be higher for mean first year than for baseline HbA1c. Each 1 % (10.9 mmol/mol) increase in mean first-year HbA1c increased 27 % the risk of major CVEs occurrence (95 % CI 11-45 %). Updating HbA1c for values obtained beyond the second year of follow-up did not improve its predictive performance. The cardiovascular protection was observed until HbA1c values lower than 6.5 % (48 mmol/mol). Moreover, the magnitude of HbA1c reduction during the first year of follow-up was predictive of better cardiovascular outcomes, independent of baseline HbA1c levels. In conclusion, better glycemic control, especially during the first year of follow-up, is determinant of better cardiovascular outcomes in high-risk patients with type 2 diabetes, without any detectable lower threshold level of HbA1c.