When there is a circumferential fracture of a cylindrical pressure vessel containing high-pressure gas, the internal pressure falls as gas escapes through the developing breach. If the break is adjacent to the end-cap, the end-cap will be accelerated by the escaping gas. Typically, the velocity of the detached end-cap is predicted using an equation proposed by Moore which assumes that the peak end-cap velocity is achieved as a result of the end-cap maintaining its initial acceleration over a distance equal to the vessel bore diameter with no subsequent acceleration. This is the approach adopted in the Nuclear Electric R3 impact assessment procedure. In order to determine the validity of the Moore equation, separate predictions are made for the end-cap velocities in the extreme cases of ＇large-mass＇ end-caps, ＇zero-mass＇ end-caps and an end-cap driven by a gas jet from a constant pressure source. These predictions suggest that the Moore equation generally overpredicts the end-cap velocity. This is confirmed by experimental data. Depending on the initial acceleration of the end-cap, its diameter relative to the vessel bore and the length of the vessel, overpredictions ranging from 10 to 150% are obtained. The data also indicate that up to 20% of the expansion energy available from the initial volume of high-pressure gas may appear as end-cap kinetic energy.