Sufficiently fast and large disruptions to the continuous price process can be detected in high frequency data as jumps. Cojumping occurs when jumps occur contemporaneously across assets. This paper assesses cojumping in the US term structure using the Cantor-Fitzgerald tick dataset of 2002-2006, and finds that the middle of the curve is more likely to cojump and the ends have greater potential for idiosyncratic jumping. What is more, cojumping is strongly associated with responses to scheduled news announcements. In instances where cojumping occurs other than in response to scheduled news the price response is smaller than with the news announcements. The results are considered over a range of sampling frequencies.