During the 1962 Johnston Island high‐altitude nuclear tests, USAELDRL operated high‐sensitivity, high‐time‐resolution magnetometers in the Pacific area and in the continental United States, Large loops of up to 100‐km2 area, spin‐resonance (metastable‐helium) magnetometers, and telluric probes were used. Shot Starfish (the July 9, 1962, explosion of about 1½ megatons at 400‐km altitude) produced very strong oscillatory signals of many minutes duration at all our stations (Hawaii, Samoa, Florida, South Carolina, New Jersey, and Maine). The signals consist of several physically different parts—a practically instantaneous broad‐band pulse (observed clearly only in the Pacific area) containing mainly higher frequencies, a strong oscillatory signal starting 1.9 seconds later simultaneously at all stations, a complex part lasting several minutes which is probably a superposition of different hydromagnetic modes, and an extremely long‐period disturbance (observed only in the Pacific area) carrying considerable energy (which we interpret as a hydrodynamic‐gravitational mode). The maximum signal amplitude (occurring within 3 to 5 seconds after the shot), when plotted versus distance from Johnston Island, results in a smoothly decreasing curve. Of the four other high‐altitude nuclear tests above Johnston Island between October 20 and November 4, 1962, which were at various heights but all considerably lower than Starfish, none gave magnetic signals clearly above the noise level at the mainland stations. Three of these events produced clear effects at Hawaii (about 1500 km to the magnetic ENE of Johnston) and at Samoa (about 3400 km magnetically south of Johnston). The signal‐amplitude ratio (Samoa to Hawaii) becomes, for sufficiently low shot heights, a pronounced function of the explosion height, indicating that the magnetic‐signal propagation becomes more and more confined to the north‐south direction as the explosion height is lowered.