Risk of substantial haemorrhage represents a critically important limitation to effective antithrombotic treatment in patients with atrial fibrillation (AF). While it is known that this risk is increased in anticoagulated patients either in the presence of anti-aggregatory drugs or concomitant renal insufficiency, there are currently few data on the potential interactions between endogenous platelet aggregability and bleeding risk. Objective: We therefore evaluated in a cohort of AF patients: (1), the putative relationship between platelet aggregability and HASBLED score; (2), the potential biochemical bases for such a relationship. Methods: Patients were included as part of SAFETY, a randomised controlled trial evaluating outpatient management of AF patients. Platelet response to ADP was evaluated via whole blood impedance aggregometry; clinical and biochemical correlates of platelet aggregation were sought via univariate and multivariate analysis. Results: Platelet aggregation correlated inversely (r=-0.220, p<0.05) with HAS-BLED score. Univariate biochemical correlates of decreased platelet aggregation were plasma concentrations of symmetric dimethylarginine (SDMA) and asymmetric dimethylarginine (ADMA). On multivariate analyses, plasma SDMA concentration (β=-0.318, p<0.01), platelet content of thioredoxin-interacting protein (Txnip, β=0.261, p<0.05) and plasma thrombospondin-1 (TSP-1, β=0.249, p<0.05) concentration were predictive of platelet ADP response. Consistent with previous reports, plasma SDMA concentrations were strongly and inversely correlated with estimated glomerular filtration rate (eGFR, r=-0.780, p<0.001). Conclusions: These data therefore suggest that (1), physiologically impaired, like pharmacologically impaired, platelet aggregability may increase bleeding risk in anticoagulated AF patients; (2), the biochemical basis for this may include impaired effects of nitric oxide (via Txnip, TSP-1) but also concomitant renal dysfunction.