The 21-aminosteroid antioxidant U-74006F (1) is being developed as an iv injectable agent for the treatment of human CNS trauma and ischemia. Because of its poor water solubility, the plasma compatibility of the parenteral formulation of 1 was evaluated using three models: (I) static solubility, (II) aggregometric, and (III) dynamic flow. The flow model was designed to mimic an iv infusion into the human antecubital vein, which was assumed to have plasma flow of 10 mL/min. Dilantin (phenytoin), the positive control, produced a precipitate in all three models from a 10% (v/v) mixture with human plasma, which approximates the in vivo ratio when the drug is infused at the recommended rate of 1 mL/min. Approximately 39% of the phenytoin dose in the flow model was retained on a downstream 3-microns filter as crystals. In comparison, the parenteral formulation of 1 produced minimal precipitate in models I and II from 40% mixtures with plasma, but higher percentages produced unstable suspensions with time-dependent precipitation. The percentage of the dose of the parenteral formulation of 1 retained on the filter in the flow model was 0.5% or less at infusion rates as high as 10 mL/min and 3% at 19 mL/min. At the 10-mL/min infusion rate, the mass of 1 retained on the filter per minute was less than 1% of the mass of phenytoin retained at the 1-mL/min infusion rate for Dilantin. The acceptable plasma compatibility of the parenteral formulation of 1 appears to be related to the solubilizing effects of plasma protein binding and pH suppression by the citric acid vehicle.