Accounts by witnesses of spring ice breakup in rivers often mention violent ice runs with extreme water speeds and rapidly rising water levels. Such events are believed to follow the release of major ice jams. To gain preliminary understanding of this problem, an attempt is made to reconstruct a partially documented ice jam release reported recently by others. The equations of the ice–water flow that occurs after the release of an ice jam are formulated. It is shown that the problem may be approximately treated as a one-dimensional, unsteady, water-only flow of total depth identical to that of the ice–water flow, and average velocity. The retarding effect of the frequently encountered intact ice cover below the jam is considered implicitly, that is, by adjusting the friction factor so as to make the predicted and observed downstream stages equal. The effects of jam length are considered next by assuming longer jams of the same maximum water depth. The duration of the surging velocities increases with jam length and so does the peak stage. Less than 2 h after the jam release the surge was arrested and a new jam formed, causing further stage increases. Present capabilities of modelling the reformation process are discussed and the major unknowns identified.