In this paper the bootstrap technique was applied to the evaluation of potential predictive bias for different ethnic groups using the Scholastic Aptitude Test (SAT) and high school ranks to predict college grade point averages (GPAs). Data of three ethnic groups were used; the total number of subjects was close to 5,000. Both conventional statistical-significance testing and the bootstrap procedure were used, and the results from the two approaches compared. For the data analyzed, when one predictor (SAT score) was used for GPA prediction, the two approaches provided somewhat different results. When two predictors (SAT score and high school rank) were used, the two approaches provided similar results. Where difference occurred, it was suggested that the results from the bootstrap approach might be more plausible. The paper demonstrated the feasibility of applying the bootstrap technique to the research in the area of predictive bias. Because the bootstrap procedure is more empirically grounded, some problems associated with conventional statistical-significance testing may be avoided. Three tables and four figures are included. (Contains 32 references.) (Author/SLD)