To determine if tryout samples typically used for item selection contribute to test bias against minority groups, item analyses were made of the California Achievement Tests using seven sub-groups of the standardization sample: Northern White Suburban, Northern Black Urban, Southern White Suburban, Southern Black Rural, Southern White Rural, Southwestern Mexican Urban and Southwestern Anglo-American Suburban. The best half of the items in each test were selected for each group. Typically about 30 percent of the items in the upper half of the distribution of item-test correlations for a group on a test did not meet this criterion with another group. By this criterion minority groups were relatively similar as were the three suburban groups. The resulting unique item tests did not correlate well with each other. Scores of minority groups were relatively better on the selected items. Thus, standard item selection procedures produce tests best suited to groups like the majority of the tryout sample and are therefore biased against other groups to some degree. This degree varies. Ways to minimize this bias need to be developed. (Author/MS)