Amchitka Island, in Alaska, was used for underground nuclear testing from 1965 to 1971. Since the test program concluded, there have been concerns about the possible release of radionuclides into the marine environment of the Aleutian Islands. The hydrogeology of islands such as Amchitka is characterized by a layer of freshwater overlying a saltwater layer, with the salinity increasing across a transition zone (TZ). Hydrogeologic modeling can provide an estimate of the timing and amount of radionuclide release from the explosions beneath Amchitka Island. This modeling is inconclusive because of a lack of information regarding subsurface structure. To address this problem, magnetotelluric (MT) data were collected on Amchitka Island in 2004. Broadband MT data were recorded on profiles passing through three explosion sites to give information about subsurface porosity and salinity. A 2D MT inversion produced models of sub-surface electrical resistivity and showed a pattern of increasing, decreasing, and increasing resistivity with depth at each test site. The depth at which resistivity begins to decrease defines the top of the TZ. The deeper increase in resistivity approximates the base of the TZ. The depths of the top and bottom of the TZ were determined as follows: Cannikin 900–2500 m; Long Shot 600–1700 m; Milrow 900–1700 m. Uncertainties were estimated for these depths. Effective porosities were also estimated and ranged from 10%–20% at the surface to 1%–3% at 3-km depth. These porosities are higher than those assumed in several hydrogeologic models, and give longer transit times from the explosion to the marine environment. Subject to the limits of the analysis, it appears that each of the cavities resulting from underground nuclear explosions is located in the TZ from fresh to saltwater. This implies shorter transit times to the marine environment than if the detonations had been located in the saltwater layer.