The relationship between item omission and item position on criterion-referenced tests in the Texas state assessment program is examined. Item statistics from the Texas Educational Assessment of Minimum Skills (TEAMS) and Texas Assessment of Basic Skills (TABS) mathematics and reading tests from 1983 through 1985 are examined for three ethnic groups and two grade levels. Sample sizes ranged from approximately 15,000 for the TEAMS field test to about 225,000 for the regular TABS testing program. Ethnic representation was approximately proportionate to that of the state. On these untimed, basic skills assessment instruments, the rate at which students omitted any item appears to increase in a regular manner with item position. The rate of increase of tiem omissions, as well as the omit rate itself, were both found to differ significantly for students from different ethnic groups. Simple regression procedures were used to fit prediction equations to these data in order to model the rate of omission as a function of item analysis. The results of this study provide information for framing test administration procedures in such a way that students from any particular ethnic group are not unfairly penalized. Categorized data are illustrated in tables and figures following the study. (Author/JAZ)