The relative toxicities of the gadolinium chelates currently available in the United States were compared when extravasated in soft tissue. The increasing use of these contrast agents in higher volumes and at faster injection rates, often with a power injector, was a principal motivation for this research. Gadopentetate dimeglumine (Magnevist), gadoteridol (ProHance), gadodiamide (Omniscan), and gadoversetamide (Optimark) were evaluated at standard concentration and compared with a control (physiologic saline) and the conventional ionic radiographic contrast medium meglumine diatrizoate (Renografin 60). Each mouse received a subcutaneous injection in the hindlimb of 0.3 mL of contrast or saline. There were 6 experimental groups, with 15 animals in each group. The individual performing the injection was blinded to the identity of the contrast agent used in each mouse. After 48 hours, the mice were killed and tissue samples obtained for histopathology. A veterinary pathologist, also blinded to the agent injected, graded the degree of damage seen on microscopic examination. Of the four MR contrast agents, gadopentetate dimeglumine caused the greatest tissue damage, and gadoteridol and gadodiamide-the two lowest osmolar agents-the least. The difference was statistically significant in terms of both inflammation (P = 0.0008 for gadoteridol, and P = 0.006 for gadodiamide) and necrosis (P = 0.0067 for gadoteridol, and P = 0.031 for gadodiamide), when these agents were compared with gadopentetate dimeglumine. In regard to the control experiments, for all three variables (necrosis, edema, and inflammation), there was no statistically significant difference between the results with gadoteridol or gadodiamide and those with saline. In terms of both edema and inflammation, the effect of gadopentetate dimeglumine, although less, could not be differentiated with any statistical significance from that of meglumine diatrizoate. Gadoversetamide, which has an osmolality between the ionic agent (gadopentetate dimeglumine) and the other two nonionic agents, caused a reaction that could not be differentiated from that seen with gadopentetate dimeglumine for both necrosis and edema. Only in the scoring of inflammation was the effect less using gadoversetamide compared to gadopentetate dimeglumine with any statistical significance (P = 0.021). The risk of tissue damage due to extravasation is not widely appreciated for the gadolinium chelates. Care should be exercised during contrast injection, to avoid inadvertent extravasation and its deleterious consequences, in particular with the two higher osmolar agents (gadopentetate dimeglumine and gadoversetamide).