The safety of gravity earth retaining structures is usually evaluated with regard to: (1) overturning about the toe, (2) sliding along the base and (3) bearing failure of the foundation. Conventional equilibrium methods are utilized in these analyses, which are performed using assumed earth loads based on simplified earth pressure theories. Recent finite element studies performed on gravity retaining walls founded on rock revealed that the use of conventional methods may lead to overly conservative results. The effects of soil-structure interaction result in a greater degree of wall stability than conventional approaches would indicate. This research examines the behavior of gravity earth retaining structures founded on soil. Two methods of analyses were used in these studies : (1) the Following Load method, which does net account for soil-structure interaction effects, and (2) the Backfill Placement method, which does account for soil-structure interaction effects. A procedure called the “Alpha Method” for 2D soil elements was developed for the purpose of improving the post-failure stress-strain behavior of the backfill and foundation soils and incorporated in the finite element program (SOILSTRUCT) utilized in the analyses. A series of analyses demonstrated the effectiveness of the Alpha Method in controlling overshoot and providing good estimates of collapse loads on wall-foundation systems. Following Load analyses indicated that walls on soil become unstable by bearing capacity rather than overturning or sliding. These results also provided the basis for modifications to Vesic’s bearing capacity theory, which extended the applicability of the theory to the conditions encountered in retaining wall problems. The Backfill Placement analyses showed that there are significant differences in behavior between walls founded on rock and walls founded on soil. These analyses also led to new insight into the factors that affect the shear forces within the backfill and which contributes to the stability of the wall. / Ph. D.