Following a dramatic increase in the seismicity and rate of ground deformation at Rabaul Caldera in September 1983, an electronic distance measuring (EDM) network was established in November--December 1983 to provide additional surveillance of the developing volcanic hazard. The network and operational procedure were designed to monitor frequently a large area and to obtain results rapidly. Initially, data reduction was carried out using the conventional refractive index technique to correct for observed atmospheric variables, but the principle of line ratios using modeled atmospheric variables showed promise for accelerating and simplifying data reduction by obviating meteorological considerations. A standard error of ¿1.31 ppm using the line ratio reduction technique favorably compares with results obtained by other researchers. Horizontal strain in the east central part of the caldera increased at monthly rates of 10--80 ppm, reaching a peak in April 1984. This peak coincided with the highest monthly rate of earthquake occurrence (13,800) and the largest monthly rate of ground uplift (approximately 80 mm). The greast amount of horizontal ground deformation recorded during the period 1983--1985 was about 400 mm. After April 1984 the strain rates declined gradually and, in relation to other components of deformation, probably resumed a precrisis level by mid-1985. Horizontal deformation in the southern part of the caldera was relatively large in comparison to the amount of ground tilt and uplift recorded there. Because of the relatively small number of reflector stations, modeling of the EDM data is somewhat inconclusive. However, a Mogi point source model provides a reasonable fit to the observed deformation in the northern part of the caldera. The point source model and simple axisymmetric models do not provide a good match to the deformation recorded in the southern part of the caldera. The dominance of horizontal deformation there could be explained by a dyke intrusion. ¿ American Geophysical Union 1988