OBJECTIVES: The relative cost-effectiveness of treatments in type 2 diabetes (T2D) can be driven by the relationship between baseline HbA1c and change in HbA1c through changes in the risk of T2D-related complications. We explore the impact of guideline-implied vs. real-world-implied estimates of baseline-adjusted changes in HbA1c on the cost-effectiveness of T2D treatments. METHODS: Recent guidelines produced by the National Institute for Health and Care Excellence (NICE) were compared to data from the EDGE study with respect to quantifying the relationship between baseline HbA1c and 1-year change in HbA1c for metformin+vildagliptin (M+V) compared to metformin+sulphonylurea (M+S); multivariate analysis of EDGE patient-level data related alternative estimates of HbA1c change to the expected lifetime incidence of T2D-related vascular complications, expressed as estimates of patient life expectancy (LE), quality-adjusted life expectancy (QALE) and cost-effectiveness, using the CORE T2D model.  RESULTS: Based on the EDGE data, estimated change in HbA1c at 1-year for M+V and M+S groups were -1.33% and -1.05%, respectively (delta: -0.28%). Applying guideline equations to the EDGE data, estimates for M+V and M+S were -0.95% and -0.99%, respectively (delta: 0.04%). In multivariate analysis, total costs were lower in M+V arm (£27,087 and £28,169) compared with  M+S arm (£28,782 and £28,988) for the EDGE and NICE comparisons respectively. This was driven by reduced incidence of complications and longer time to insulin intensification in the M+V arm. The NICE approach for estimating HbA1c change predicted a smaller difference in predicted change in QALE (M+V vs. M+S: 0.31 vs. 0.30 years: delta 0.01) compared with the EDGE data (M+V vs. M+S: 0.41 vs. 0.31 years: delta 0.10). CONCLUSIONS: The NICE T2D guidelines may underestimate the health economic value of M+V. The applicability of clinical guidelines to clinical practice should be considered when evaluating the cost-effectiveness of treatment options in T2D.