In the early nineteenth century, fragmentary weather observing networks were established in portions of the United States, primarily to constitute a climatological history. The invention of the telegraph in 1833 and its availability to the public in 1845 eventually led to a nationwide observation network with the formation of a national weather service in 1870 within the Signal Service of the U.S. Army. While Lt. J. P. Finley of the Signal Service was developing techniques for the prediction of tornadoes, others, as Prof. Henry A. Hazen (also of the Signal Service) and the New England Meteorological Society, were studying the thunderstorm in hope of developing predictors for the forecasting of these events. The thunderstorm studies were carried over to the civilian U.S. Weather. Bureau in 1891, but interest in thunderstorm and tornado research wanted as a ban on the use of the word “tornado” in forecasts, inherited from the Signal Corps, was carried into the twentieth century. The growth of the aviation industry after World War I forced the bureau to accept the theories of the Norwegian school, as forecasts of thunderstorms and upper-air conditions (winds, ceilings, visibilities, etc.) were mandated by law. World War II brought greater changes with the development of the radiosonde and radar, and more emphasis was directed to the study of thunderstorms. Early in the war, the U.S. Weather Bureau was requested to develop warning networks in the vicinity of munitions plants to report the approach of thunderstorms in order to evacuate the plant workers to safety lest lightning struck. These networks spread to military installations, and eventually were organized for selected cities in tornado-prone areas of the central plains and the Missouri River valley. The Thunderstorm Project, conceived in 1943, began field operations in 1947 to obtain a detailed description of all phases of the thunderstorm, not necessarily to differentiate thunderstorm intensity. At the same time, Fawbush and Miller were in the process of developing criteria to make this differentiation in order to predict severe thunderstorms and tornadoes. In 1949, the Weather Bureau began to develop procedures to forecast severe local storms, which culminated in the formation of a crash research program called the Tornado Project. The success of experimental tornado forecast by the project led to the establishment of the Severe Weather Unit (SWU) in May 1952 for the prediction of severe local storms nationwide. Later, it was renamed the Severe Local Storms Center (SELS), now a component of the National Severe Storms Forecast Center (NSSFC). Forecast development and research from 1950 to 1963 is presented with primary attention given to the development of techniques for forecasting severe local storms and the emergence of SELS as a viable forecast unit.