It was argued that the language used in multiple choice achievement test items should be no more complex than is necessary to test the examinee's knowledge of the subject matter. Excessive complexity might constitute a source of bias against people with limited verbal skills. A test of this hypothesis was made by comparing a simplified language version of STEP social studies test items against an "original language form" with junior and senior high school students in Oakland, California. About two-thirds of the students were Black. Mean reading comprehension scores fell at about the 30th percentile on national norms. Test results on both forms appeared to be internally consistent and in accord with general expectations for this student sample. There was, however, no support for the notion that the simplified language test version would yield different results. Test difficulty and relationships with reading comprehension scores and social studies grades were essentially the same for both forms.