We test the suitability of short-occupation differential and absolute positioning methods using data from a 13 station global positioning system (GPS) network that spans the Santa Ana Volcano and Coatepeque caldera of western El Salvador for monitoring intereruptive activity and tectonic movements near these potentially hazardous features. Data spanning a 1 yr period from 12 GPS benchmarks located 1.9 km to 9.7 km from a continuous reference GPS station were processed with Trimble differential processing software to determine the repeatabilities and hence precisions of the differential station coordinates. For observation sessions spanning 20 min, the coordinates of the benchmark closest to our reference station are repeatable to within 4–6 mm in the horizontal component of the baseline between the two sites and 12–13 mm for the vertical component. In contrast, the horizontal and vertical repeatabilities at the site farthest from the reference station are 7–11 mm and 30–34 mm, respectively. These results suggest that any horizontal movement of the benchmarks larger than ~10 mm relative to the reference site, or vertical movement larger than 10–30 mm (depending on the baseline distance) should be detectable. Five sites adjacent to and within Coatepeque caldera moved upward at rates of 180–470 mm/yr relative to the reference site from February to July 2009. In contrast, no uplift pattern during this period was observed for the other network sites, suggesting an uplift source beneath the caldera. No increase in microseismicity coincided with the transient inflation event, and no other possibly corroborating observations from the caldera are available. The cause of the uplift is thus unknown. Our results suggest that differential GPS with occupation times of 20 min or more is a useful monitoring tool at subtropical volcanoes and calderas for networks with baselines that are shorter than ~10 km. Absolute positioning as an alternative GPS processing method gives precisions of ~10 mm (95%) in the absolute station coordinate estimates for occupations as short as 3 h, i.e., sufficiently precise for monitoring volcano deformation. This new monitoring strategy, whereby one dual-frequency receiver is used for short benchmark occupations without any need for a reference station, is simpler than that required for differential monitoring and minimizes tropospheric water vapor as a source of noise in station position estimates.