VLF signals propagating in the Earth-ionosphere waveguide are used to probe the heated nighttime D region over three U. S. Navy very low frequency (VLF, 3-30 kHz) transmitters. Ionospheric cooling and heating are observed when a transmitter turns off and on in the course of normal operations. Heating by the 24.0-kHz NAA transmitter in Cutler, Maine, (1000 kW radiated power) was observed by this method in 41 of 52 off/on episodes during December 1992, increasing the amplitude and retarding the phase of the 21.4-kHz NSS probe wave propagating from Annapolis, Maryland, to Gander, Newfoundland, by as much as 0.84 dB and 5.3 ¿, respectively. In 6 of these 41 episodes, the amplitude of the 28.5-kHz NAU probe wave propagating from Puerto Rico to Gander was also perturbed by as much as 0.29 dB. The latter observations were unexpected due to the >770 km distance between NAA and the NAU-Gander great circle path. Heating by the NSS (21.4 kHz, 265 kW) and NLK (24.8 kHz, 850 kW) transmitters was observed serendipitously in data from earlier measurements of the amplitudes of VLF signals propagating in the Earth-ionosphere waveguide. A three-dimensional model of wave absorption and electron heating in a magnetized, weakly ionized plasma is used to calculate the extent and shaped of the collision frequency (i.e., electron temperature) enhancement above a VLF transmitter. The enhancements are annular, with a geomagnetic north-south asymmetry and a radius at the outer half-maximum of the collision frequency enhancement of about 150 km. Heating by the NAA transmitter is predicted to increase the nighttime D region electron temperature by as much as a factor of 3. The calculated changes in the D region conductivity are used in a three-dimensional model of propagation in the Earth-ionosphere waveguide to predict the effect of the heated patch on a subionospheric VLF probe wave. The range of predicted scattered field amplitudes is in general consistent with the observed signal perturbations. Discrepancies in the predictions are attributed to lack of knowledge of the D region electron density profile along with the probe wave great circle paths. ¿ American Geophysical Union 1994