A total of 174 mandibles from 5 morphologically distinct samples (African-Americans, Ankara Native-Americans, Caucasians, Inuit, and Pacific Islanders) were analyzed, using 17 landmarks (points of comparison) from the skeletal surface of each mandible. All of the landmarks are identifiable on both adult and sub-adult specimens. Three-dimensional coordinates of the landmarks were collected with a Microscribe 3DX desktop digitizing system. The digitized coordinates of the mandibles were superimposed by means of generalized Procrustes analysis. The results showed significant differences in mandible measurements among all five racial groups. The study also assessed the practicality of using this racial identification technique for incomplete skeletal remains, since sub-adult skeletal material often does not survive burial intact. This was done by determining the potential for identifying race from partial mandibular material. Group distinction by race was achieved at an accuracy of 67 percent for the mandibular corpus and 73 percent for the ramus, using a sample from three racial groups. These results show a reduction in accuracy compared to those achieved from the complete mandible, but the accuracy levels still remained high, suggesting that an attempt at racial determination can be made with partial mandibular remains. 7 tables, 4 figures, and 27 references