Thirty large mesio-occlusodistal cavities with their margins totally in enamel were restored using the three-sited light-curing technique in five different ways: group I = Mark I version of a light-reflecting wedge; groups 2 and 3 = i nterproximal curing with transparent, nonreflecting wedges; groups 4 and 5, Mark II version of the light-reflectin g wedge with increased resiliency. In groups 3 and 5, the light that was not conducted by the transparent wedges was shielded by a piece of aluminum foil. It was found that, when used directly and unshielded, both Mark I and Mark II versions of the light-reflecting wedge induced significantly better marginal adaptation than the transparent, nonreflecting wedge, especially gingivoproximally. With both the nonreflecting and reflecting wedges, the shielding had no effect on the overall percentage of excellent margin. However, gingivoproximally, the shielding significantly improv ed the marginal quality in the nonreflecting wedge groups, leading to the conclusion that the first increment should be cured as throughly as possible from a gingivoproximal direction, preferably indirectly via light-conducting wedges. Direct irradiation from a lingual or buccal direction is less favorable.