This paper presents the results of full-scale loading tests performed on 54 passive anchors and 4 group anchored footings grouted to various lengths at several sites in Korea. The test results, the failure mechanisms as well as uplift capacities of rock anchors depend mostly on rock type and quality, embedded fixed length, properties of the discontinuities, and the strength of rebar. Anchors in poor quality rocks generally fail along the grout/rock interfaces when their depths are very shallow (a fixed length of less than 1 m). However, even in such poor rocks, we can induce a more favorable mode of rock pull-up failure by increasing the fixed length of the anchors. On the other hand, anchors in good quality rocks show rock pull-up failures with high uplift resistance even when they are embedded at a shallow depth. Laboratory test results revealed that a form of progressive failure usually occurs starting near the upper surface of the grout, and then progresses downward. The ultimate tendon-grout bond strength was measured from $18{sim}25%$ of unconfined compressive strength of grout. One of the important findings from these tests is that the measured strains along the corrosion protection sheath were so small that practically the reduction of bond strength by the presence of sheath would be negligible. Based on test results, the main parameters governing the uplift capacity of the rock anchor system were determined. By evaluation of the ultimate uplift capacity of anchor foundations in a wide range of in situ rock masses, rock classification suitable for a transmission tower foundation was developed. Finally, a very simple and economical design procedure is proposed for rock anchor foundations subjected to uplift tensile loads.