When runoff begins, uplift pressures develop on the underside of the shore fast cover. Analysis shows that longitudinal cracks should form soon after runoff starts. Two cracks are normally predicted, subdividing the cover into a main central part and two side strips. Where ice thickness is large or the channel is narrow, a single mid-channel crack is predicted. Field observations support the theory. With increasing flow, the central portion of the cover may detach and become subject to transverse fractures. The latter could result from bending on vertical or horizontal planes. Vertical bending fracture requires extreme, wavelike slopes such as might prevail briefly during jam releases. Flow shear and the meandering planform of rivers cause horizontal bending manifested in commonly observed transverse crack patterns. The resulting separate ice sheets will be set in motion if there is enough room on the water surface between the river boundaries, thus initiating the breakup. This mechanism leads to a breakup forecasting criterion, called the boundary constraint, that explains past empirical findings and identifies the factors influencing various empirical coefficients. Another type of breakup is caused by ice-jam releases and formation of breaking fronts whereby the ice cover is reduced to rubble before it moves. Little is known about the motion of breaking fronts but relevant field observations are reviewed. Key words: bending, breakup, crack, cover, fracture, ice, onset, release, river, surge.