In order to examine whether different results are obtained from two sociometric techniques for determining children's peer preferences in desegregated schools, two studies were conducted. The techniques compared were (1) the traditional peer nomination method, in which children list a few classmates whom they consider friends; and (2) the roster and rating method, in which each child rates all classmates on an interval scale. The first study involved meta analysis of existing data. Results of research using the two sociometric methods were compared and it was found that peer nomination studies tend to find more same-race preferences than roster and rating studies. The second study involved field research. Sixth graders at a newly desegregated school were asked to show their peer preferences on both types of measures. Results indicated that the peer nomination method results in larger differences in preferences for same-race peers over peers of a different race. These findings suggest that peer nomination techniques, which restrict the number of choices a child can make and thus encourage the naming of best friends, should be used to assess close friendships; roster and rating methods may be more appropriately used to assess more general intergroup acceptance. (Author/MJL)