In 2019, the EAT-Lancet Commission proposed a planetary, healthy, and reference diet, designed to reduce the global health–environmental double burden of current dietary patterns. We aim to investigate the association between the EAT-Lancet reference diet and type 2 diabetes (T2D) risk. Methods This observational and prospective web-based study was conducted using data from 88 964 adults participating to the French NutriNet-Santé cohort 2009–22 (78.8% female; mean (SD) baseline age of participants was 43.8 (14.5) years). Dietary intakes were assessed at least by three repeated 24 hour-dietary records (mean = 6.3 (2.8)). The endpoint was incident T2D. The main exposure was the EAT-Lancet diet index (ELD-I) modeled as a continuous variable and sex-specific quintiles (Qs). Hazard ratios (HRs) and confidence intervals (CIs) were assessed by Cox proportional hazards models adjusted for potential confounders. The body mass index (BMI) role as a potential mediating factor in the relationship was also investigated. Results During follow-up (median = 8.79 years), 812 incident T2D cases occurred. The ELD-I ranged from −192 to 429 points with a median (interquartile range [IQR]) score of 42.8 (29.1) points. For each 25.7-point increase in ELD-I score (1SD), the risk of T2D is reduced by 11% (HR1SD: 0.89; 95% CI: 0.82–0.96; P = .003). The association studied was mediated up to 61% by BMI. Conclusions In this large cohort study, an increase in adherence to the EAT-Lancet reference diet was associated with a decreased risk of T2D beyond BMI. These results support previous findings and highlights the interest of the EAT-Lancet reference diet as a healthy and sustainable diet.