A fusible plug is a threaded cylinder of metal, usually bronze, brass or gunmetal, with a tapered hole drilled completely through its length. This hole is sealed with a metal of low melting point that flows away if a predetermined high temperature is reached. The initial use of the fusible plug was as a safety precaution against low water levels in steam engine boilers, but later applications extended its use to other closed vessels, such as air conditioning systems and tanks for transporting corrosive or liquefied petroleum gases.

A fusible plug operates as a safety valve when dangerous temperatures, rather than dangerous pressures, are reached in a closed vessel. In steam boilers the fusible plug is screwed into the crown sheet (the top plate) of the firebox, typically extending about 1 in (25 mm) into the water space above it. Its purpose is to act as a last-resort safety device in the event of the water level falling dangerously low: when the top of the plug is out of the water it overheats, the low-melting-point core melts away and the resulting blast of steam into the firebox warns the operators before the top of the firebox itself runs completely dry, which could result in a boiler explosion. The flue gases in a steam engine firebox can reach 1,000 °F (538 °C), a temperature that copper, from which most fireboxes were made, softens to the point it can no longer sustain boiler pressure and explode if water is not quickly added to the boiler and the fire damped down or extinguished. The hole through the plug is too small to greatly reduce steam pressure and the small amount of water, if any, that passes through is not sufficient to lower the fire.

The device was invented in 1803 by Richard Trevithick, the proponent of high-pressure (as opposed to atmospheric) steam engines, in consequence of an explosion in one of his new boilers. His detractors were eager to denounce the whole concept of high-pressure steam, but Trevithick proved that the accident happened because his fireman had neglected to keep the boiler full of water. He publicised his invention widely, without patent, to counter these criticisms.

Experiments conducted by the Franklin Institute, Boston, in the 1830s had initially cast doubt on the practice of adding water as soon as the escape of steam through the device was noted. A steam boiler was fitted with a small observation window of glass and heated beyond its normal operating temperature with the water level below the top of the firebox. When water was added it was found that the pressure rose suddenly and the observation glass shattered. The report concluded that the high temperature of the metal had vaporised the added water too quickly and that an explosion was the inevitable result.

It was not until 1852 that this assumption was challenged: Thomas Redmond, one of the Institute's inspectors, specifically ruled out this theory in his investigation into the boiler explosion on the steamship Redstone on the Ohio River on 3 April that year. A 1907 investigation in Wales came to a similar conclusion: a steam locomotive belonging to the Rhymney Railway was inadvertently sent out with its safety valves wrongly assembled. The pressure in the boiler built up to the extent that the injectors failed; the crown sheet became uncovered, was weakened by the heat of the fire and violently blew apart. The investigation, led by Colonel Druitt of the Railway Inspectorate, dismissed the theory that the enginemen had succeeded in starting the injectors and that the sudden flood of cold water had caused such a generation of steam that the boiler burst. He quoted the results of experiments by the Manchester Steam Users' Association, a national boiler certification and insurance body, that proved that the weight of copper present (considered with its specific heat) was insufficient to generate enough steam to raise the boiler pressure at all. Indeed, the addition of cold water caused the pressure to fall. From then on it was accepted that the correct action in the event of the operation of the fusible plug was to add water.

The simple solid plug is filled with a slug of low-melting-point alloy. When this melts, it first melts as a narrow channel through the plug. Steam and water immediately begins to escape through this. As the water will have a maximum temperature of 410 °F (210 °C), lower than tin's melting point of 410 °F, this water jet may act to freeze the plug. While water continues to escape from the plug, the plug may fail to melt completely and so only a minor jet of steam is noticed, which may be overlooked.

To avoid this, the cored fusible plug was developed in the 1860s to give a wide opening as soon as the alloy softens. This has a solid brass or bronze centre, soldered into place by a thick layer of the low-melting-point alloy. When overheated, the plug does not release any steam or water until the alloy melts sufficiently to release the centre plug. The plug now fails dramatically, opening its entire bore immediately. This full-bore jet is then more likely to be noticed.

A drawback to the device was found on 7 March 1948, when the firebox crown sheet of Princess Alexandra, a Coronation Pacific of the London, Midland and Scottish Railway, failed while hauling a passenger train from Glasgow to London. Enquiries established that both water gauges were defective and on a journey earlier that day one or both of the fusible plugs had melted, but this had gone unnoticed by the engine crew because of the strong draught carrying the escaping steam away from them.