The associations between arginine-based metabolites and incident type 2 diabetes (T2D) are unknown. We employed a case-cohort design, nested within the PREDIMED trial, to examine six plasma metabolites (arginine, citrulline, ornithine, asymmetric dimethylarginine [ADMA], symmetric dimethylarginine [SDMA] and N-monomethyl-l-arginine [NMMA]) among 892 individuals (251 cases) for associations with incident T2D and insulin resistance. Weighted Cox models with robust variance were used. The 1-year changes in arginine (adjusted hazard ratio [HR] per SD 0.68, 95% confidence interval [CI] 0.49, 0.95; Q4 vs. Q1 0.46, 95% CI 0.21, 1.04; P trend = 0.02) and arginine/ADMA ratio (adjusted HR per SD 0.73, 95% CI 0.51, 1.04; Q4 vs. Q1 0.52, 95% CI 0.22, 1.25; P trend = 0.04) were associated with a lower risk of T2D. Positive changes of citrulline and ornithine, and negative changes in SDMA and arginine/(ornithine + citrulline) were associated with concurrent 1-year changes in homeostatic model assessment of insulin resistance. Individuals in the low-fat-diet group had a higher risk of T2D for 1-year changes in NMMA than individuals in Mediterranean-diet groups (P interaction = 0.02). We conclude that arginine bioavailability is important in T2D pathophysiology.