OBJECTIVE: The extravasation detection accessory (EDA) is designed for use during contrast-enhanced CT studies performed with a power injector. The EDA detects the changes in soft-tissue impedance that occur with enhanced extravasation and halts the further infusion of contrast material via a feedback circuit to the injector. We tested the sensitivity of this device in a model of contrast extravasation. MATERIALS AND METHODS: Study subjects had an extravasation of 5% dextrose in water (nonionic contrast equivalent) in one arm and 0.9% sodium chloride solution (ionic contrast equivalent) in the other. An EDA was placed over the site of infusion and connected to a power injector. Injections were performed at 0.25 ml/sec (n = 40), 2.5 ml/sec (n = 62), or 5 ml/sec (n = 20). RESULTS: At infusion rates of 2.5 and 5 ml/sec, the device halted the injector in every subject after an average volume of 12.5 +/- 1.6 ml was delivered. At 0.25 ml/sec, the device failed to halt the injector in 11 of 20 events. After reprogramming the algorithm, 10 more subjects were tested at the lowest injection rate. The device halted 18 of 20 extravasation events with an average volume of 3.7 +/- 0.5 ml. CONCLUSION: In our model of contrast extravasation, the EDA halted a power injector with reliability and reproducibility before a large volume of contrast material was delivered. The sensitivity of the device approached, but did not reach, 100%. This device may serve to diminish the morbidity of extravasation events.