Autologous blood transfusion is performance enhancing, and detection of abuse remains a major challenge in sports. To address this issue, investigators transfused a group of men with either blood or saline, and evaluated whether metabolites collected in urine before and after the transfusion could indicate which transfusion was administered. More than 2,500 different urine compounds were analyzed and a machine learning analytical approach was applied. The results indicated that the model allows identification of ~66% of the blood-transfused individuals for up to three days after the transfusion. The research team also discovered that plasticizers in urine are strong indicators of blood doping. The plastic compounds probably originated from the plastic bag that the blood was stored in before transfusion. Furthermore, and importantly, investigators highlight that no other urine metabolites appeared to be of value for the detection of a blood transfusion.