This data set contains traverse measurements used to infer the mass balance of central parts of the ice sheet upslope from the traverse, and results from this work have been published in Thomas et al., [1998] and 2000, with comparisons of these results with those from repeat altimetry in Thomas et al. (2001). Ongoing projects using these data include: Collaboration with J. Bamber to increase the spatial resolution of our mass-balance estimates in order to identify causes for major imbalance in parts of southern Greenland, and use of the ice velocities to infer basal conditions along the traverse. During 1993-1997 a line of stakes about 30 km apart was established completely around the ice sheet approximately along the 2000 m elevation contour except in the south west, where it was higher because of high mountains, nunataks, and crevasses [Thomas et al., 1998 and 2000]. Each station was precisely surveyed by GPS, with resurvey one or two years later, to yield estimates of ice velocity. Station locations were marked by aluminum poles approximately 2 meters long, inserted close to vertically into the snow surface to a depth of about 1 meter. A second and, in areas of very high accumulation, a third pole was added to the lower pole using connecting sections and hose clamps. At each station, a GPS receiver was set atop the lower pole, by attaching a short length of aluminum rod to the base of the receiver and inserting this into the marker pole. Most stations were installed and surveyed via Twin Otter aircraft, during brief (about 10 minute) stops for station installation, and GPS mounting etc. The GPS receivers were left to operate for a minimum of 40 minutes, and more commonly a few hours. Up to 10 stations were visited during a one-day flight from a nearby coastal station. The GPS data from each day were processed by a point-positioning analysis for the longest-occupied station, with differential solutions for the others relative to this. Errors on resulting horizontal locations were about +/- 2 cm for the longest-occupied station, increasing to about +/- 10 cm for the shortest occupancy, with vertical errors approximately double these values. Additional errors were introduced by tilting of the marker poles during the 1-2 year interim between surveys, increasing the uncertainty by perhaps 5 cm. Consequently, derived velocities are accurate to better than +/- 0.2 m/yr. The data were acquired with support from NASA's Polar Research Program, and users requesting the data are asked to provide a brief description of their application of the data. This will help provide NASA with an assessment of the scientific benefits resulting from their support of such efforts.