Introduction: Robust predictive measures for Type 2 diabetes (T2D) vascular complications to facilitate early intervention strategies are needed. Methods: Fasting plasma from 363 Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) trial participants (mean±SD) age: 61.3±6.4 yrs, 5.6±5.4 yrs T2D, HbA1c: 6.7±1.2% (49.7±13.4 mmol/mol), were collected at baseline and stored (-80°C) until quantitative proteomics analysis of tryptic peptides. Protein signatures associated with future micro- or macrovascular complications (100 and 54 subjects, respectively) were identified using a random forest approach, adjusted for treatment allocation. Exhaustive search using logistic regression was then carried out to optimise models. Pathway analysis was carried out using the NIH functional annotation tool, DAVID Bioinformatics. Resources. Results: Fifty targets were selected: 14 were associated with microvascular, 13 with macrovascular complications and 23 were common for both. Pathway analysis revealed six main pathways: cholesterol metabolism, platelet activation, complement and coagulation, focal cell adhesion, actin cytoskeleton and hypertrophic cardiomyopathy. In logistic regression, the optimal model for prediction of microvascular complications with overall p<0.0001, AUC=0.74 contained Cadherin-5, Clusterin, Ceruloplasmin, Glutathione peroxidase 3 and Inter-alpha-trypsin inhibitor heavy chain H3. For macrovascular complications the model contained Zinc-alpha-2-glycoprotein and Coagulation factor V (p<0.0001, AUC=0.72). For any micro- or macro-vascular complications, the model contained Cholinesterase, Ceruloplasmin, Mannose-binding protein C and Coagulation factor V (p<0.0001, AUC=0.70). Conclusion: Proteomic signatures represent a promising first step towards identifying protein signatures associated with T2D complications. Future research will aim at finding therapeutic targets in the metabolic pathways identified by those proteins.